PeopleSoft 8.8 Analytic Forecasting PeopleBook

Transcription

1 PeopleSoft 8.8 Analytic Forecasting PeopleBook December 2002

2 PeopleSoft 8.8 Analytic Forecasting PeopleBook SKU EPM88ANF-B 1202 PeopleBooks Contributors: Teams from PeopleSoft Product Documentation and Development. Copyright PeopleSoft, Inc. All rights reserved. Printed in the United States. All material contained in this documentation is proprietary and confidential to PeopleSoft, Inc. ("PeopleSoft"), protected by copyright laws and subject to the nondisclosure provisions of the applicable PeopleSoft agreement. No part of this documentation may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, including, but not limited to, electronic, graphic, mechanical, photocopying, recording, or otherwise without the prior written permission of PeopleSoft. This documentation is subject to change without notice, and PeopleSoft does not warrant that the material contained in this documentation is free of errors. Any errors found in this document should be reported to PeopleSoft in writing. The copyrighted software that accompanies this document is licensed for use only in strict accordance with the applicable license agreement which should be read carefully as it governs the terms of use of the software and this document, including the disclosure thereof. PeopleSoft, PeopleTools, PS/nVision, PeopleCode, PeopleBooks, PeopleTalk, and Vantive are registered trademarks, and Pure Internet Architecture, Intelligent Context Manager, and The Real-Time Enterprise are trademarks of PeopleSoft, Inc. All other company and product names may be trademarks of their respective owners. The information contained herein is subject to change without notice. Open Source Disclosure This product includes software developed by the Apache Software Foundation ( Copyright (c) The Apache Software Foundation. All rights reserved. THIS SOFTWARE IS PROVIDED AS IS AND ANY EXPRESSED OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE APACHE SOFTWARE FOUNDATION OR ITS CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. PeopleSoft takes no responsibility for its use or distribution of any open source or shareware software or documentation and disclaims any and all liability or damages resulting from use of said software or documentation.

3 Contents General Preface About This PeopleBook......vii PeopleSoft Application Prerequisites...vii PeopleSoft Application Fundamentals...vii Related Documentation......viii Obtaining Documentation Updates...viii Ordering Printed Documentation...viii Typographical Conventions and Visual Cues......ix Typographical Conventions...ix Visual Cues...x Comments and Suggestions......xi Common Elements in These PeopleBooks......xi Preface PeopleSoft Analytic Forecasting Preface......xiii PeopleSoft Enterprise Warehouse......xiii Other Related PeopleBooks......xiii Chapter 1 Getting Started with PeopleSoft Analytic Forecasting Overview...1 Setting Up Metadata...1 Implementing PeopleSoft Analytic Forecasting Setting Up PeopleSoft Analytic Forecasting Tables...3 Chapter 2 Introducing PeopleSoft Analytic Forecasting Understanding Forecasting Understanding Analytic Forecasting Components...6 How Forecasting Works...7 Establishing Forecast Cycles...9 Using Multidimensionality Using Forecast-Free Models PeopleSoft Proprietary and Confidential iii

4 Contents Using History to Create Datasets Using Forecasting Methods Understanding Mathematical Formulations Solving Multiple Linear Regression Problems...14 Computing Multiple Linear Regression Parameters...15 Analyzing Statistical Errors for Multiple Linear Regression...16 Chapter 3 Welcome to PeopleSoft Analytic Forecasting Analytic Forecasting Home Page Pages Used to Navigate the Application...19 Navigating Through the Application Chapter 4 Setting Up Forecasting Understanding Forecasting Rules Defining Forecast Options and Reviewing Statistical Method Data Pages Used to Set Up Options...22 Setting Up Forecasting Options...22 Viewing Statistical Methods...22 Setting Up Seasonality Groups Page Used to Set Up Seasonality Groups...23 Defining Seasonality Groups...23 Setting Up Control Groups Pages Used to Set Up Control Groups...25 Setting Up Control Groups...25 Specifying the Scheduled Growth Forecast Profile...26 Specifying the Exponential Smoothing Forecast Profile...28 Using the Control Group Clone Utility Page Used to Clone Control Groups...29 Copying a Control Group...29 Defining Datasets Page Used to Define Datasets...29 Setting Up Datasets...29 Creating Forecast Definitions Pages Used to Create Forecast Definitions...32 Defining Your Forecast Definition Parameters...32 Establishing the Dataset Code...34 iv PeopleSoft Proprietary and Confidential

5 Contents Specifying an Independent Variable for a Multivariate Forecast...35 Specifying the Seasonality Group...35 Specifying the Control Group...36 Scheduling the Forecast Execution Cycle and Entering Notes...36 Reviewing Your Forecast Definitions...36 Defining Forecast Groups Page Used to Define Forecast Groups...37 Setting Up a Forecast Group...37 Using the Forecast Clone Utility Page Used to Clone Forecasts...39 Copying a Forecast Definition...39 Defining Forecast Elements Pages Used to Define Forecast Elements...40 Viewing and Creating Key Values...40 Defining Performance Parameters...41 Viewing Statistical Parameters...42 Viewing Regression Parameters...42 Viewing Smoothing Parameters...43 Viewing Seasonality Factors...43 Chapter 5 Processing, Reviewing, and Adjusting Forecasts Overview Using Engines and Jobstreams Engines Used in Analytic Forecasting...46 Using Jobstreams...47 Using Analytic Forecasting Run Control Pages...47 Running Jobs from the Run Control Pages...48 Reviewing Forecast Jobstream Messages...48 Tables Used for Forecasting Output Using Multicurrency Functionality Importing Historical Data Working with Forecast Processing Viewing Charts and Printing Results...51 Adjusting Your Forecast Output Pages Used to Adjust Output...53 Reviewing and Adjusting Forecast and Historical Data...53 Viewing and Updating Seasonality Data...55 Viewing and Processing Exceptions...56 PeopleSoft Proprietary and Confidential v

6 Contents Reviewing Forecast Data with Crystal Reports Appendix A PeopleSoft Analytic Forecasting Reports PeopleSoft Analytic Forecasting Reports: A to Z Glossary of PeopleSoft Terms...61 Index...73 vi PeopleSoft Proprietary and Confidential

7 About This PeopleBook PeopleBooks provide you with the information that you need to implement and use PeopleSoft applications. This preface discusses: PeopleSoft application prerequisites. PeopleSoft application fundamentals. Related documentation. Typographical elements and visual cues. Comments and suggestions. Commonelements in PeopleBooks. Note. PeopleBooks document only page elements that require additional explanation. If a page element is not documented with the process or task in which it is used, then either it requires no additional explanation or it is documented with common elements for the section, chapter, PeopleBook, or product line. Elements that are common to all PeopleSoft applications are defined in this preface. PeopleSoft Application Prerequisites To benefit fully from the information that is covered in these books, you should have a basic understanding of how to use PeopleSoft applications. See Using PeopleSoft Applications. You might also want to complete at least one PeopleSoft introductory training course. You should be familiar with navigating the system and adding, updating, and deleting information by using PeopleSoft windows, menus, and pages. You should also be comfortable using the World Wide Web and the Microsoft Windows or Windows NT graphical user interface. These books do not review navigation and other basics. They present the information that you need to use the system and implement your PeopleSoft applications most effectively. PeopleSoft Application Fundamentals Each application PeopleBook provides implementation and processing information for your PeopleSoft database. However, additional, essential information describing the setup and design of your system appears in a companion volume of documentation called the application fundamentals PeopleBook. Each PeopleSoft product line has its own version of this documentation. PeopleSoft Proprietary and Confidential vii

8 General Preface The application fundamentals PeopleBook consists of important topics that apply to many or all PeopleSoft applications across a product line. Whether you are implementing a single application, some combination of applications within the product line, or the entire product line, you should be familiar with the contents of this central PeopleBook. It is the starting point for fundamentals, such as setting up control tables and administering security. Related Documentation This section discusses how to: Obtain documentation updates. Order printed documentation. Obtaining Documentation Updates You can find updates and additional documentation for this release, as well as previous releases, on the PeopleSoft Customer Connection Website. Through the Documentation section of PeopleSoft Customer Connection, you can download files to add to your PeopleBook Library. You ll find a variety of useful and timely materials, including updates to the full PeopleSoft documentation that is delivered on your PeopleBooks CD-ROM. Important! Before you upgrade, you must check PeopleSoft Customer Connection for updates to the upgrade instructions. PeopleSoft continually posts updates as the upgrade process is refined. See Also PeopleSoft Customer Connection Website, Ordering Printed Documentation You can order printed, bound volumes of the complete PeopleSoft documentation that is delivered on your PeopleBooks CD-ROM. PeopleSoft makes printed documentation available for each major release shortly after the software is shipped. Customers and partners can order printed PeopleSoft documentation by using any of these methods: Web Telephone Web From the Documentation section of the PeopleSoft Customer Connection Website, access the PeopleSoft Press Website under the Ordering PeopleBooks topic. The PeopleSoft Press Website is a joint venture between PeopleSoft and Consolidated Publications Incorporated (CPI), the book print vendor. Use a credit card, money order, cashier s check, or purchase order to place your order. viii PeopleSoft Proprietary and Confidential

9 General Preface Telephone Contact CPI at Send to CPI at See Also PeopleSoft Customer Connection Website, Typographical Conventions and Visual Cues This section discusses: Typographical conventions. Visual cues. Typographical Conventions The following table contains the typographical conventions that are used in PeopleBooks: Typographical Convention or Visual Cue Description Bold Indicates PeopleCode function names, method names, language constructs, and PeopleCode reserved words that must be included literally in the function call. Italics Indicates field values, emphasis, and PeopleSoft or other book-length publication titles. In PeopleCode syntax, italic items are placeholders for arguments that your program must supply. We also use italics when we refer to words as words or letters as letters, as in the following: Enter the number 0, not the letter O. KEY+KEY Indicates a key combination action. For example, a plus sign (+) between keys means that you must hold down the first key while you press the second key. For ALT+W, hold down the ALT key while you press W. Monospace font Indicates a PeopleCode program or other code example. (quotation marks) Indicate chapter titles in cross-references and words that are used differently from their intended meanings. PeopleSoft Proprietary and Confidential ix

10 General Preface Typographical Convention or Visual Cue Description... (ellipses) Indicate that the preceding item or series can be repeated any number of times in PeopleCode syntax. { } (curly braces) Indicate a choice between two options in PeopleCode syntax. Options are separated by a pipe ( ). [ ] (square brackets) Indicate optional items in PeopleCode syntax. & (ampersand) When placed before a parameter in PeopleCode syntax, an ampersand indicates that the parameter is an already instantiated object. Ampersands also precede all PeopleCode variables. (ISO) Information that applies to a specific country, to the U.S. federal government, or to the education and government market, is preceded by a three-letter code in parentheses. The code for the U.S. federal government is USF; the code for education and government is E&G, and the country codes from the International Standards Organization are used for specific countries. Here is an example: (GER) If you re administering German employees, German law requires you to indicate special nationality and citizenship information for German workers using nationality codes established by the German DEUEV Directive. Cross-references PeopleBooks provide cross-references either below the heading See Also or on a separate line preceded by the word See. Cross-references lead to other documentation that is pertinent to the immediately preceding documentation. Visual Cues PeopleBooks contain the following visual cues. Notes Notes indicate information that you should pay particular attention to as you work with the PeopleSoft system. Note. Example of a note. x PeopleSoft Proprietary and Confidential

11 General Preface A note that is preceded by Important! is crucial and includes information that concerns what you must do for the system to function properly. Important! Example of an important note. Warnings Warnings indicate crucial configuration considerations. Pay close attention to warning messages. Warning! Example of a warning. Comments and Suggestions Your comments are important to us. We encourage you to tell us what you like, or what you would like to see changed about PeopleBooks and other PeopleSoft reference and training materials. Please send your suggestions to: PeopleSoft Product Documentation Manager PeopleSoft, Inc Hacienda Drive Pleasanton, CA Or send comments to doc@peoplesoft.com. While we cannot guarantee to answer every message, we will pay careful attention to your comments and suggestions. Common Elements in These PeopleBooks As of Date Business Unit Description Effective Date Once, Always, and Don t Run The last date for which a report or process includes data. An ID that represents a high-level organization of business information. You can use a business unit to define regional or departmental units within a larger organization. Enterupto30charactersoftext. The date on which a table row becomes effective; the date that an action begins. For example, to close out a ledger on June 30, the effective date for the ledger closing would be July 1. This date also determines when you can view and change the information. Pages or panels and batch processes that use the information use the current row. Select Once to run the request the next time the batch process runs. After the batch process runs, the process frequency is automatically set to Don t Run. Select Always to run the request every time the batch process runs. Select Don t Run to ignore the request when the batch process runs. PeopleSoft Proprietary and Confidential xi

12 General Preface Report Manager Process Monitor Run Request ID User ID SetID Short Description Click to access the Report List page, where you can view report content, check the status of a report, and see content detail messages (which show you a description of the report and the distribution list). Click to access the Process List page, where you can view the status of submitted process requests. Click to access the Process Scheduler request page, where you can specify the location where a process or job runs and the process output format. An ID that represents a set of selection criteria for a report or process. An ID that represents the person who generates a transaction. An ID that represents a set of control table information, or TableSets. TableSets enable you to share control table information and processing options among business units. The goal is to minimize redundant data and system maintenance tasks. When you assign a setid to a record group in a business unit, you indicate that all of the tables in the record group are shared between that business unit and any other business unit that also assigns that setid to that record group. For example, you can define a group of common job codes that are shared between several business units. Each business unit that shares the job codes is assigned the same setid for that record group. Enter up to 15 characters of text. See Also Using PeopleSoft Applications PeopleSoft Process Scheduler xii PeopleSoft Proprietary and Confidential

13 PeopleSoft Analytic Forecasting Preface This preface discusses: Enterprise Warehouse PeopleBooks. Other Related PeopleBooks. PeopleSoft Enterprise Warehouse Additional, essential information describing the setup and design of your system appears in a companion volume of documentation called PeopleSoft 8.8 Enterprise Warehouse PeopleBook. Note. OneormorepagesinPeopleSoftAnalytic Forecasting operate in deferred processing mode. Deferred processing is described in the preface of the PeopleSoft 8.8 Enterprise Warehouse PeopleBook. Other Related PeopleBooks This PeopleBook provides detailed implementation and setup information common to all, or more than one, of the following PeopleSoft applications: PeopleSoft 8.8 Activity-Based Management PeopleSoft 8.8 HRMS Warehouse PeopleSoft 8.8 Workforce Planning PeopleSoft 8.8 Workforce Rewards PeopleSoft 8.8 Enteprise Scorecard See Also PeopleSoft 8.8 Enterprise Warehouse PeopleBook PeopleSoft Proprietary and Confidential xiii

14 Preface xiv PeopleSoft Proprietary and Confidential

15 CHAPTER 1 Getting Started with PeopleSoft Analytic Forecasting This chapter provides an overview of PeopleSoft Analytic Forecasting and discusses how to implement Analytic Forecasting. Overview Implementing PeopleSoft Analytic Forecasting requires that you fully understand your current and proposed business processes, as well as the cost and profitability metrics that have the greatest influence on your organization s goals and objectives. This knowledge, combined with PeopleSoft Analytic Forecasting, will keep your planning system focused. The implementation process includes setting up specifications in the PeopleSoft Enterprise Warehouse application that will enable PeopleSoft Analytic Forecasting to perform as designed. Setting Up Metadata PeopleSoft Analytic Forecasting uses metadata to describe the data stored in PeopleSoft Enterprise Warehouse. Analytic Forecasting uses five types of metadata: TableMaps, DataMaps, filters, constraints, and datasets. You typically set these up when you set up Enterprise Warehouse. The following diagram shows the metadata creation hierarchy. PeopleSoft Proprietary and Confidential 1

16 Getting Started with PeopleSoft Analytic Forecasting Chapter 1 Dataset Constraint Filter DataMap TableMap Hierarchy of metadata creation Metadata is important in PeopleSoft Analytic Forecasting because it establishes the attributes of the historical source data. Metadata establishes the rules around which the forecast import process operates, by pointing to the primary tables in PeopleSoft Enterprise Warehouse from which it obtains source data. PeopleSoft Analytic Forecasting provides the following view tables, and the metadata that is based on these view tables. Primary Record Description of View Sample Dataset FI_INSTR_VW Financial instruments used for Financial Services Industry purposes INSTR_ORIG (no constraint) REVENUE_VW Revenue REVENUE_VW PF_STAT_VW Statistics PF_STAT_VW CALC_OBJ_QT_VW Calculated object quantity CALCOBJ_QT PF_LEDGER_VW2 PF Ledger PFLED_FCST Note. You can establish constraints under the SetID SHARE, or with whatever SetID you wish. TableMaps and DataMaps are not keyed by SetID. Understanding Constraints and Filters PeopleSoft Analytic Forecasting uses filters and constraints to select subsets of available data. Filters enable you to select the specific data that you want to forecast. For instance, to construct a forecast for fixed-rate mortgages, you would create a filter for only the product ID associated with fixed mortgages. If fixed-rate mortgages were your only selection criteria, then you would simply attach a single filter to a constraint. 2 PeopleSoft Proprietary and Confidential

17 Chapter 1 Getting Started with PeopleSoft Analytic Forecasting Constraints enable you to group several filters together. For example, if you want to forecast sales for checking accounts that are offered over the Internet, you must first filter each dimension, then combine the filters in a constraint and, finally, reference that constraint in the forecast dataset. Using filters and constraints in this manner enables you to limit the values that the Import engine will consider when it creates elements. You must create your own filters for all of the view tables. Implementing PeopleSoft Analytic Forecasting This section describes the tables you define to implement PeopleSoft Analytic Forecasting. The following steps are listed in suggested implementation sequence. The actual sequence by which you implement and use the application may differ. In the planning phases of your implementation, take advantage of all PeopleSoft sources of information, including installation guides, ETL mapping reports, and data models. A complete list of these resources is in the preface of the PeopleSoft Enterprise Warehouse PeopleBook with information about where to find the most current version of each. You can use the following table to set up the PeopleSoft Analytic Forecasting application. The tasks listed take you through the individual steps for setting up and running forecasts. See Also PeopleSoft 8.8 Enterprise Warehouse PeopleBook, PeopleSoft Enterprise Warehouse Preface Setting Up PeopleSoft Analytic Forecasting Tables Follow these steps to set up PeopleSoft Analytic Forecasting. Step 1. Set up PeopleSoft Enterprise Warehouse, including business rules, general options, PF and general ledger business units, ledgers and templates, and ledger mapper templates. 2. Import historical data into the PeopleSoft Enterprise Warehouse ODS (Operational Data Store) using an ETL tool. Reference PeopleSoft 8.8 Enterprise Warehouse PeopleBook PeopleSoft 8.8 Enterprise Warehouse PeopleBook 3. Set up metadata. PeopleSoft 8.8 Enterprise Warehouse PeopleBook 4. Setupglobaloptions. Chapter 4, Setting Up Forecasting, Defining Forecast Options and Reviewing Statistical Method Data, page Setupseasonalitygroups. Chapter 4, Setting Up Forecasting, Setting Up Seasonality Groups, page Setupcontrol groups. Chapter 4, Setting Up Forecasting, Setting Up Control Groups, page Setupdatasets. Chapter 4, Setting Up Forecasting, Defining Datasets, page 29 PeopleSoft Proprietary and Confidential 3

18 Getting Started with PeopleSoft Analytic Forecasting Chapter 1 Step Reference 8. Set up forecast definitions. Chapter 4, Setting Up Forecasting, Creating Forecast Definitions, page Set up forecast groups. Chapter 4, Setting Up Forecasting, Defining Forecast Groups, page Define forecast elements. Chapter 4, Setting Up Forecasting, Defining Forecast Elements, page Set up jobstreams. PeopleSoft 8.8 Enterprise Warehouse PeopleBook 12. Run PeopleSoftAnalytic Forecasting. Chapter 5, Processing, Reviewing, and Adjusting Forecasts, Running Jobs from the Run Control Pages, page 48 Chapter 3, Welcome to PeopleSoft Analytic Forecasting, Analytic Forecasting Home Page, page 19 For individual jobs, or for jobstreams. 13. Review forecasting results. Chapter 5, Processing, Reviewing, and Adjusting Forecasts, Working with Forecast Processing, page 50 Chapter 3, Welcome to PeopleSoft Analytic Forecasting, Analytic Forecasting Home Page, page 19 4 PeopleSoft Proprietary and Confidential

19 CHAPTER 2 Introducing PeopleSoft Analytic Forecasting This chapter provides an overview of forecasting and discusses: PeopleSoft Analytic Forecasting components. How forecasting works. Establishing forecast cycles. Using multidimensionality. Using forecast-free models. Using history to create datasets. Using forecasting methods. Understanding mathematical formulations. Understanding Forecasting The main challenge of forecasting is the sheer number of forecasts that must be produced every period. For example, if your organization has 3,000 products, three channels, and two business units, this would amount to 18,000 forecasts each month. PeopleSoft Analytic Forecasting addresses this particular challenge by enabling you to create forecasts for your business requirements. The system uses a primary table view, in conjunction with PeopleSoft Enterprise Warehouse metadata, to import pertinent history data, which it then couples with a control group and a seasonality group. The control group specifies the statistical method, and the seasonality group defines the parameters used to derive a forecast. PeopleSoft Analytic Forecasting enables you to import external forecasts as well as create your own forecasts. If you are using PeopleSoft Enterprise Warehouse as a data warehouse, you can import external forecasts. To import an external forecast, you must: 1. Establish an ETL map from your source system to the target forecast output table. 2. Create a forecast definition with the Imported check box selected, that will be used to associate the imported forecast data with key values on the forecast output tables, through the ETL map. PeopleSoft Proprietary and Confidential 5

20 Introducing PeopleSoft Analytic Forecasting Chapter 2 Understanding Analytic Forecasting Components PeopleSoft Analytic Forecasting enables you to forecast effectively by using datasets, control groups, and seasonality groups. These three components exist separately within PeopleSoft Analytic Forecasting, to enable you to pick and choose precisely which ones you want to use within a given forecast definition. For each forecast definition, you define a specific dataset, at least one control group, and an optional seasonality group (or groups). You can also define explicit forecast elements to create forward-looking data as part of your forecast. Finally, you can group related forecast definitions into forecast groups. The following illustration shows how the forecasting components relate to each other. Forecast Definition Elements A single combination of dimensions (for example, product and channel ) Control Group Dataset one instance Seasonality Group one instance Forecasting Method Forecasting Options optional Quantity dates Control Group Seasonal behavior Dataset Seasonality Group Relationships among forecasting components The following provides a brief overview of each Analytic Forecasting component. Datasets Datasets are the source for forecasting and comprise history data used according to quantity, date, and key dimension values. The data can be imported through the ODS/ETL process. Control Groups Control groups represent groupings of particular properties of the forecast, such as what degree of accuracy to expect and what statistical method to apply. You set up control groups on the Control Group page. 6 PeopleSoft Proprietary and Confidential

21 Chapter 2 Introducing PeopleSoft Analytic Forecasting Seasonality Groups Seasonality groups capture default seasonality values to apply to a forecast. Seasonality groups also associate similar elements within a forecast during the computation of seasonality factors from history data. You set up seasonality groups on the Seasonality Group page. Forecast Elements Forecast elements represent the specific combination of dimensions, such as product and channel, that occur within a dataset. For example, let s say your organization markets three different soda products: lemon, lime, and berry. Let s also say that you operate within two channels: wholesale and retail. A forecast element, then, is simply an occurrence of one of each: for example, retail lemon soda, or wholesale berry soda. Most forecast elements are created automatically by PeopleSoft Analytic Forecasting. The system is able to track each occurrence in the dataset. You can, optionally, set up explicit, manual forecast elements if you need forward-looking data, by using the Forecast Elements component. For example, let s say your organization had not previously sold berry soda wholesale, but wants to forecast the revenue opportunities which may exist. In this case, because you have no history data, you have to create that data manually. However, because there may be as few as one or as many as five thousand elements in a forecast definition, it is generally more efficient to have PeopleSoft Analytic Forecasting create the elements for you. Forecast Groups You can create and group together several forecast definitions to form a forecast group. Instead of having to process individual forecasts that may be too numerous to effectively manage, you can use forecast groups to associate and group related forecast definitions for processing. Running Analytic Forecasting PeopleSoft Analytic Forecasting is a cyclical process that estimates or calculates data which then serves as a prediction of upcoming conditions. All forecasting jobs are based on a scenario containing a forecast group. The system processes all rules in the forecast definitions within that group. After the Forecast engine runs, you, as the forecaster, are able to determine whether the forecast values are reasonable. You can update production forecasts regularly (monthly, quarterly, or at any other interval) using the Process Scheduler. You can create simulation forecasts as needed or process them regularly, similar to the way you manage production forecasts. How Forecasting Works Statistical forecasting also called historical forecasting occurs in two main phases: pre-forecasting and forecasting, as represented by the following diagram. PeopleSoft Proprietary and Confidential 7

22 Introducing PeopleSoft Analytic Forecasting Chapter 2 Pre-Forecasting Audit Rules Setup Adjust History (if necessary) Import Current Data Find and Reset Inaccurate Forecasts Forecasting Compute Seasonality Factors Recompute All Forecasts Adjust Forecasts Forecasting flow Pre-Forecasting You pre-forecast first. To do so, you: 1. Audit your rules setup. 2. Adjust your history data if necessary. 3. Import current data to add to existing history data. 4. Assess your forecast accuracy. Sometimes, the forecasting accuracy of the statistical parameters diminishes as more history data is imported. If your accuracy diminishes, you must reset the forecast element. Resetting simply marks the forecast so that the system recreates the statistical parameters governing the forecast from the history data. The forecast methods remain the same. Provided that the forecast is sufficiently accurate, the system continues to use the same statistical for forecasting each time the forecast is recomputed. Forecasting Actual forecasting involves: 1. Computing seasonality factors. 2. Recomputing all forecasts and saving with the new as of date. 3. Reviewing the forecasts. 8 PeopleSoft Proprietary and Confidential

23 Chapter 2 Introducing PeopleSoft Analytic Forecasting 4. Adjusting the forecasts until they are satisfactory. Forecasters review the forecasts and enter adjustments that cannot be inferred statistically. For example, let s say your organization has a promotional campaign next quarter that is expected to boost volume for certain products over several weeks. In such a situation, you would only apply adjustments to one forecast element at a time. Forecasting generally occurs as a largely automated batch process. In each period, you import new history numbers, the active period (the first period for which a forecast is computed) moves ahead, and statistical forecasts are refitted, if necessary. One or more forecasters review the forecast and make adjustments. The final forecast can be used, for example, by data marts, and by other applications, such as PeopleSoft Activity-Based Management. Establishing Forecast Cycles Because you forecast periodically on a monthly basis for example you must specify the active year and period for each forecast. The active year and period denote the first period for which the system produces a forecast. You select one date within the activeyearandactiveperiodtobetheas of date. This is typically the first or last date in the period under consideration. Each new forecast version is keyed by the as of date. This enables you to compare previous forecast versions to the current one. Completing the entire forecast cycle may require a few hours or a few days, depending on the volume and level of activity. The following table lists specific events that occur during a forecast cycle. User Task System Task Update the forecast schedule, specifying a new as of date. Import new values into the history. Review exceptions and adjust any history values that are outliers. Check for excess variance between the old forecast and the latest actuals and reset poorly performing forecasts. Adjust forecasts based on non-statistical business intelligence. Recompute all statistical forecasts. Adjust history values that are adversely affecting the statistics. Recompute the reset forecasts. Mark the forecast complete. Reapply adjustments to the new forecast. You establish the active year and active period for the forecasting cycle on the Schedule Status page. This page enables you to initiate each new forecasting cycle. PeopleSoft Proprietary and Confidential 9

24 Introducing PeopleSoft Analytic Forecasting Chapter 2 Using Multidimensionality Multidimensionality is a feature that enables PeopleSoft Analytic Forecasting to incorporate multiple dimensions into forecasts automatically. PeopleSoft Analytic Forecasting produces multidimensional forecasts at the detail level, based on forecast elements. The forecast for a product can be generated across several dimensions; for example, the product forecast by channels, by customers, and by departments. To limit complexity, Analytic Forecasting designates a small number of dimensions (for example, products, channels, and customers) to participate in the forecast. Any dimensions you don t select on the dataset are aggregated. For example, let s say you have a record containing the following dimensions: Account Customer Channel Product Department If you choose all of these except Channel on the dataset, then the detail for Channel will be aggregated for forecasting purposes. Aggregate dimensions also include the forecast version, time, and unit of measure (volume, revenue, cost, and so forth). The forecasting process manages these dimensions, but not as true dimensions. In some circumstances, you ll want to forecast more than one quantity associated with a specified product; for example, new loan origination and total balances for a loan category. These are two different forecasts, even though they are attached to the same product, and may use different methods and have different control parameters and statuses. Time is also a dimension, and forecasting can, in principle, occur at several levels in the time dimension. To limit complexity, however, forecasting is done only within one detail calendar the calendar that is assigned to the scenario. Using Forecast-Free Models Sometimes historical values are used in computing quantities that enter into other calculations but are not themselves forecast. For example, the beta of a stock portfolio measures the historical change in value relative to that of its market. Historical values relating the prices of the portfolio and its universe are needed to estimate the beta, but forecasts for future values of the beta are not normally required. Aside from the lack of a forecast, however, a beta is similar to a forecast quantity in that it is a multidimensional quantity, its model is revised periodically, it depends statistically upon the historical values of other quantities, and it comes with error estimates. In such cases you would specify PeopleSoft Analytic Forecasting to forecast historical values for 0 (zero) periods forward. The model is created, fitted, maintained, and checked against reasonability conditions, but no forecast is computed; therefore, no forecast accuracy exceptions occur. In the previous paragraph s example, the quantities of interest are the slope of the regression (beta) and the basis value (alpha). 10 PeopleSoft Proprietary and Confidential

25 Chapter 2 Introducing PeopleSoft Analytic Forecasting Note. To use a forecast-free model, specify 0 (zero) in the Total Forecast Periods field on the Forecast Definition - Definition page or in the Forecast Periods To Use field on the Forecast Element - View Keys page, where you define the number of forecast periods and history periods. See Also Chapter 4, Setting Up Forecasting, Defining Forecast Elements, page 39 Chapter 4, Setting Up Forecasting, Understanding Forecasting Rules, page 21 Using History to Create Datasets PeopleSoft Analytic Forecasting enables you to specify which data to use from PeopleSoft Enterprise Warehouse. On the dataset setup page, you specify which history data you want to use to create a dataset, what dimensions to use in breaking the forecasts into individual elements, and what date field to use to measure time. For example, your dataset may contain the fields (columns) and values (rows) shown in the following table: Product ID Department ID Ship Date Balance Amount P A February 3, 2000 USD 20 P B May 12, 2000 USD 20 Q A May 6, 2000 USD 50 Q A May 12, 2000 USD 20 The aggregation type Group By is used with PeopleSoft Analytical Forecasting to compile values of like dimensions. For example, if a dataset within a forecast definition specifies a Group By aggregation type on Product ID and Department ID, PeopleSoft Analytic Forecasting automatically uses the two group keys (Product ID and Department ID) to collect data into an element it can use. Furthermore, PeopleSoft Analytic Forecasting uses the designated date field (Ship Date, in this example) to group rows into fiscal years and accounting periods, according to the calendar; then all rows with ship dates in May 2000 will be grouped and summarized under the 2000/5 accounting period. One field, (Balance Amount, in this example) is designated as the quantity to forecast. You define how to aggregate all the quantities occurring within the same accounting period. For example, you may want to sum all balances in the same month to give a total balance. The function (sum, count, average, and so forth) that is used to compute the aggregation is called an aggregation type. Assuming that the system uses the aggregation type sum, the system collects the dataset as history as shown in the following table: PeopleSoft Proprietary and Confidential 11

26 Introducing PeopleSoft Analytic Forecasting Chapter 2 Product ID Department ID Fiscal Year and Period Balance Amount P B 2000/5 USD 20 Q A 2000/5 USD 70 This table now shows the history as two distinct forecast elements (the distinct combinations of product ID, department ID, and fiscal year and period). Each has its own forecast, and each can be managed individually within a forecast definition. Note. You need history data for linear regression control groups, but history is optional for scheduled growth control groups. Using Forecasting Methods With Analytic Forecasting, we provide three different sets of methods for forecasting: Univariate (scheduled growth). Multivariate (linear regression). Exponential smoothing (Holt-Winter s). In addition, forecasts that are computed by other products, such as PeopleSoft Demand Planning, can be imported into PeopleSoft Analytic Forecasting and used exactly as if they were computed by one of the three listed methods. The following table outlines the techniques that support the forecasting methods: Forecasting Method Technique How it Works Univariate Scheduled growth Uses recorded history or an explicit base value to obtain basic forecasts quickly, using simple methods that are based on annual percentage growth. Univariate also supports explicit forecast values. Scheduled growth can use seasonal methods to add patterns of seasonal variation to the forecast. 12 PeopleSoft Proprietary and Confidential

27 Chapter 2 Introducing PeopleSoft Analytic Forecasting Forecasting Method Technique How it Works Multivariate Linear regression Uses recorded history and other forecasts to determine causal factors and the relationships to the target value. Can also use seasonal variation to smooth the history data and add patterns of seasonal variation to the forecast. Exponential smoothing Holt-Winter s This is a very popular scheme to produce a smoothed Time Series. Whereas in Single Moving Averages the past observations are weighted equally, Exponential Smoothing forecasts future values by assigning exponentially decreasing weights as the observation get older. In other words, recent observations are given relatively more weight in forecasting than the older observations. The Holt-Winter s method, which includes seasonality, allows you to choose a specific model for best representing the history data in the least squares sense. The Holt-Winter s method then uses a three updating equations scheme, each with a constant that ranges from 0to1topredictthefuturevalues. The equations are intended to give more weight to recent observations and less weight to observations further in the past. Univariate Method Univariate forecasting methods derive a base value against which to forecast, and then apply a growth percentage to it. A univariate base value can be derived in one of two ways: From the most recent period history value. From a value that you explicitly define (history is not required). If both an explicit and a history period value exist, the history value always takes precedence in batch processing. If you want to include a forecast value that is explicitly stated (not derived from a growth percent and base value), you can do so with a univariate method. PeopleSoft Analytic Forecasting supports scheduled growth as the technique for univariate forecasting. PeopleSoft Proprietary and Confidential 13

28 Introducing PeopleSoft Analytic Forecasting Chapter 2 Multivariate Method Multivariate forecasting methods use both the recorded history for the target value and the history and forecasts for other variables (causal factors), to infer not only a forecast for the target value, but also a functional relationship between the causal factors and the target value. Multivariate techniques are especially important in simulations where the future of a causal factor such as interest rates can assume many sequences, depending on the simulation scenario. Multivariate forecasts introduce another requirement into the forecast cycle; you must manage the sequence in which you compute the forecasts. You must do this because the forecasts for causal factors have to be available to forecast values that are dependent upon them. PeopleSoft Analytic Forecasting supports linear regression as the technique for multivariate forecasting. Exponential Smoothing Exponential smoothing forecast methods are useful for routine short-term forecasting of a large number of time series. For example, in inventory control applications, you may need to forecast the sales of a large number of products on a regular basis (monthly, weekly, or daily). Because of their simplicity, exponential smoothing techniques can be used very effectively in such situations. In exponential smoothing, however, you must determine (or estimate) one or more smoothing parameters, and these choices determine the weights assigned to the observations. A major disadvantage of this technique, however, is that there is no formal model building procedure. The choice often depends on the interpretation of a plot of the time series. The choice of a particular method means that you must decide whetheratrendexists. Youmustalsodecide whether seasonal variation exists. If seasonal variation does exist, you must decide if it is increasing or constant (multiplicative or additive). PeopleSoft Analytic Forecasting supports the Holt-Winter s exponential smoothing as the technique for exponential smoothing. Generating Exceptions and Adjustments The system generates exceptions automatically for the forecasting methods. Exceptions point to data input into the system, or computed by the system, that is too large or too small according to criteria you define. The system also maintains your adjustments to the history and forecast values. The system automatically reapplies these persistent adjustments when the forecast is updated in subsequent cycles, so that you do not have to keep readjusting your forecasts. You can view forecasting techniques on the Forecasting Methods page; however, it is on the Control Group page that you must specify a technique for your forecast. Understanding Mathematical Formulations PeopleSoft Analytic Forecasting uses different types of mathematical formulas to compute forecasts. These formulas are discussed in the following sections. Solving Multiple Linear Regression Problems The relationship between the response (or dependent) variable y and n-1 independent variables sometimes takes the form. 14 PeopleSoft Proprietary and Confidential

29 Chapter 2 Introducing PeopleSoft Analytic Forecasting With a set of m( > n-1) observations of dependent variable at its corresponding independent variables for i=1,,m. The coefficients, are usually chosen to minimize or equivalently, in its matrix-vector form, to minimize (1) where A is an m nmatrixandy is an m-vector such that (2) and with the n-vector Computing Multiple Linear Regression Parameters The solution of (1) is well known by where (3) PeopleSoft Proprietary and Confidential 15

30 Introducing PeopleSoft Analytic Forecasting Chapter 2 is the n m pseudo inverse of matrix A such that is the singular value decomposition of A. Note that U and V are m mandn n orthogonal matrices (that is, and ), respectively; is the m n nonnegative diagonal matrix and is the n m (nonnegative) diagonal matrix with diagonal elements given by Analyzing Statistical Errors for Multiple Linear Regression The following standard statistical quantities for analyzing the multiple linear regression results can be found in most regression analysis books. The least squares fit for y:. The sum squares of residuals of y:. The degree of freedom for y:. The variance of fitted y:. The variance-covariance matrix:. The standard error of 16 PeopleSoft Proprietary and Confidential

31 Chapter 2 Introducing PeopleSoft Analytic Forecasting (2). The predicted y at :. The mean vector : The standard error of a predicted. y: PeopleSoft Proprietary and Confidential 17

32 Introducing PeopleSoft Analytic Forecasting Chapter 2 18 PeopleSoft Proprietary and Confidential

33 CHAPTER 3 Welcome to PeopleSoft Analytic Forecasting This chapter provides an overview of the Analytic Forecasting home page and discusses how to navigate through the application. Analytic Forecasting Home Page The launching point for the PeopleSoft Analytic Forecasting application is the Analytic Forecasting home page. From here you can access the various areas of the application by using the specific links. Each of these links takes you to a separate home page from which you can access the individual pages of the application. The Analytic Forecasting home page provides a launching point to the following main pages: The Forecast Processing page from which you can access pages to run individual forecasting engines to process your forecast. The Forecast Setup page from which you can access all the forecasting rules setup pages. The Forecast Results page from which you can access pages to graphically view your historical and forecast results, as well as make adjustments to those values. TheForecastReporting page from which you can access all the Analytic Forecasting report run controls. Pages Used to Navigate the Application Page Name Object Name Navigation Usage Analytic Forecasting Home AF_INTRO_HOME Analytic Forecasts, Analytic Forecasting Home Forecast Processing AF_PROCESS_HOME Analytic Forecasts, Analytic Forecasting Home, Forecast Processing Forecast Setup AF_RULES_HOME Analytic Forecasts, Analytic Forecasting Home, Set Up Forecast Rules Forecast Results AF_RESULT_HOME Analytic Forecasts, Analytic Forecasting Home, Forecast Results and Charting Forecast Reporting AF_REPORTS_HOME Analytic Forecasts, Analytic Forecasting Home, Review Forecast Reports Access the various areas of the Analytic Forecasting application. Access all pages to process your forecasts. Access all pages for setting up your Forecasting Rules. Access all pages for reviewing your forecast results and exceptions. Access the run control pages for running all the delivered Analytic Forecasting reports. PeopleSoft Proprietary and Confidential 19

34 Welcome to PeopleSoft Analytic Forecasting Chapter 3 Navigating Through the Application Access the Analytic Forecasting Home page to access any of the other four main areas by clicking the appropriate link. Analytic Forecasting Home page Once you re on any of the four main pages of the application, for example the Forecast Processing page, an Analytic Forecasting Home link at the bottom of the page returns you to the Analytic Forecasting Home page. You access specific pages of the application from the appropriate main page by clicking the link for that page or component. Once you are done with the page and have saved your changes, you canclickthe Previous Page link to return to the specific main page or the Analytic Forecasting HomelinktoreturntotheAnalyticForecastingHomepage. 20 PeopleSoft Proprietary and Confidential

35 CHAPTER 4 Setting Up Forecasting This chapter provides an overview of forecasting rules and discusses how to: Define forecast options and reviewing statistical method data. Set up seasonality groups. Set up control groups. Use the Control Group Clone Utility. Define datasets for forecasting. Create forecast definitions. Defineforecast groups. Use the Forecast Clone Utility. Define forecast elements. Understanding Forecasting Rules Once you have completed your PeopleSoft Enterprise Warehouse setup, you are ready to define your forecasting rules. You can access the setup pages by clicking the Define Forecasting Rules link on the Analytic Forecasting home page. This will take you to the Forecasting Rules Home page from which you can define forecasting rules and the associated parameters, seasonality factors, source data, and statistical methods. Defining Forecast Options and Reviewing Statistical Method Data As a first step in setting up your forecast rules, we recommend you do the following: On the Forecast Options page, if you re using regression, specify the number of regression variables to be used in a particular forecast. On the Forecasting Methods page, view the statistical method data delivered with your system. Important! It is critical that you review the demo data because it will help you determine what you must set up for your own forecasts. PeopleSoft Proprietary and Confidential 21

36 Setting Up Forecasting Chapter 4 PagesUsedtoSetUpOptions Page Name Object Name Navigation Usage Forecasting Options AF_OPTS_PNL Analytic Forecasts, Analytic Forecasting Home, Set Up Forecast Rules, Forecasting Options Forecasting Methods AF_METH_PNL Analytic Forecasts, Analytic Forecasting Home, Set Up Forecast Rules, View Statistical Methods Setting Up Forecasting Options Access the Forecasting Options page. In the Max Independent Variables field, specify the number of regression variables to be used in a given forecast. Specifying a set number prevents you from running large regression forecasts that could inhibit system functionality. Warning! Specifying more than five independent variables may significantly tax memory and adversely affect processing speed. Specify the maximum number of regression variables that can be used in a particular forecast. View the data delivered with the system. Viewing Statistical Methods Access the Forecasting Methods page. Forecasting Methods page Review the following display only fields for the information you need to set up your forecasts. Minimum History Periods Multivariate Seasonal The minimum number of history periods required for the statistical method to work. Generally, this is the most useful field on the page because it identifies the amount of data needed for forecasting. When selected, indicates that this method is a multivariate method. Otherwise, it is a univariate method. When selected, indicates that this method can account for seasonal variations making externally applied seasonal adjustments unnecessary. 22 PeopleSoft Proprietary and Confidential

37 Chapter 4 Setting Up Forecasting Calculates Standard Error Parameter Has Min (has minimum) Min (minimum) Has Max (has maximum) Max (maximum) Optimized User Set Opt From (optional from) Opt To (optional to) Opt Step (optional step) Default Character Value When selected, indicates that this method can calculate standard errors statistically. Shows the parameters for this method. When selected, indicates there is a minimal number of history periods shown in the Min field. The minimal number of history periods. Values appear only if the Has Min check box is selected. When selected, indicates there is a minimal number of history periods shown in the Max field. The maximum number of history periods. Values appear only if the Has Max check box is selected. When selected, shows the optimal value of the parameter, if one exists. When selected, indicates there are user-defined values for this parameter. The lowest parameter value the system will consider when searching for the optimal value. The highest parameter value the system will consider when searching for the optimal value. Indicates when the system will step at a certain amount from lowest to highest while searching. The default value for non-numeric parameters. Setting Up Seasonality Groups If you are using seasonality factors in your forecast, you must set them up on the Seasonality Group page. The seasonality groups you define are linked to your forecast during setup of a forecast rule. Page Used to Set Up Seasonality Groups Page Name Object Name Navigation Usage Seasonality Group AF_SEASGP_PNL Analytic Forecasts, Analytic Forecasting Home, Set Up Forecast Rules, Define Seasonality Group Defining Seasonality Groups Access the Seasonality Group page by entering a SetID and seasonality group ID. Place forecast elements into a seasonality group, specify seasonality group characteristics, and apply seasonality to forecast output. PeopleSoft Proprietary and Confidential 23

38 Setting Up Forecasting Chapter 4 Seasonality Group page To set up a seasonality group: 1. Enter a Description for the group. 2. In the Calendar group box, enter the Calendar ID of the calendar you want to use for this seasonality group. The system prompts the calendar ID from the Calendar Builder page. The seasonality group uses it as a basis for calculating the seasons. 3. In the Periods Per Seasonal Cycle field, enter the number of accounting periods you want to define per seasonal cycle. The calendar ID limits the options. The number of periods per cycle must repeat evenly into a year relative to the calendar ID. For example, a quarterly calendar cannot have five periods per seasonal cycle (that is, more periods than available in a single year). Similarly, a monthly calendar cannot have eight periods per seasonal cycle, because eight does not repeat evenly into 12 months (one year). 4. Select the Freeze Seasonality Factors check box if you want the default seasonality factors to display in the Seasonality Group page after you have adjusted the seasonality. If you do not select this check box, the system bases the calculation of seasonality factors in the Seasonality page on the history values of all active elements associated with the seasonality group. 5. In the Default Seasonality Factors group box, for each Accounting Period enter the Seasonality Factor. The system generates the accounting periods based on the Periods Per Seasonal Cycle setting. The sum of the seasonality factors you enter must equal one (1). 24 PeopleSoft Proprietary and Confidential

39 Chapter 4 Setting Up Forecasting Setting Up Control Groups Control groups define the statistical method used in a given forecast calculation. You select the statistical method and define specific parameters associated with that method. You then associate the control groups to the forecast during the setup of a forecast rule. You set up control groups using the Define Control Groups component. Each forecast definition must have at least one control group. Using the Define Control Groups component, you can: Define the control group and statistical method. Specifythescheduled growth forecast profile, if applicable. Specify the exponential smoothing forecast profile, if applicable. See Also Pages Used to Set Up Control Groups Page Name Object Name Navigation Usage Control Group AF_CTRLGP_PNL Analytic Forecasts, Analytic Forecasting Home, Set Up Forecast Rules, Define Control Group Scheduled Growth AF_CGPROF_PNL Select the Scheduled Growth tab. Exponential Smoothing AF_CGESPROF_PNL Select the Exponential Smoothing tab. Define a control group and the options associated with a control group, and to specify the statistical method. Add additional parameters, such as annual percentage growth and standard error percentage for univariate forecasts. Specify the forecast profile and define additional parameters for forecasts using exponential smoothing. Setting Up Control Groups Access the Control Group page. PeopleSoft Proprietary and Confidential 25

40 Setting Up Forecasting Chapter 4 Control Group page To set up a control group: 1. Select the Statistical Method you want to use in your forecast: EXPSMOOTH MULTVARREG SCHEDGROW An exponential smoothing method that forecasts future values using history data with a user-specified smoothing model. A multivariate linear regression technique that uses recorded history and other forecasts. A univariate method that forecasts a quantity by specifying explicit values and subsequent annual percentage changes in quantity. If you select the SCHEDGROW method, you must enter a numeric starting point for the forecast in the Base Value field. 2. Specify the number of standard errors expected or tolerated for the forecast in the Threshold # Standard Errors field. The Forecast Accuracy engine will report any deviations exceeding this value as exceptions. 3. Select the Deseasonalize check box to remove the seasonality inherent in the history data before the forecast is run and to reapply the seasonality computed forecast. You determine the seasonal pattern to use on the Forecast Definition - Seasonality Group page. If you do not select this check box, forecasts will be generated using only the seasonality inherent in the history data. Specifying the Scheduled Growth Forecast Profile If you selected a statistical method of SCHEDGROW on the Control Group page, access the Scheduled Growth page. 26 PeopleSoft Proprietary and Confidential

41 Chapter 4 Setting Up Forecasting Scheduled Growth page To specify the scheduled growth forecast profile, enter the following Scheduled Growth Forecast Profile data: From Year From Period To Year To Period Sched Growth (scheduled growth) Forecast Value Annual % Growth (annual percentage growth) Target Balance Standard Error % (standard error percentage) First year to which the rule on this row applies. First period to which the rule on this row applies. Last year to which the rule on this row applies. Last period to which the rule on this row applies. Select this check box to ignore scheduled growth calculation using forecast value and annual percentage growth, and to instead use the target balance value for the periods and years specified. The data will then populate the Forecast output table (AF_FCST_F00) as a constant value for the forecast value. Numeric value to represent the forecast value. Annual percentage growth value for the forecast value. The value to which the percent growth applies is selected, in order, as follows: 1. The forecast value from the last year and period of the preceding rule; or 2. The last history value for the forecast element; or 3. A base value specified for the forecast element; or 4. A base value specified for the control group. (Optional) Enter a forecasted target balance amount. Value that prompts an exception to be created if the number of standard errors exceeds this amount. Note. When you have a scheduled growth forecast which starts prior to the periods defined in the control group, the forecasting engines will use the base value as the forecast value and apply one percent (1 %) as the standard error percentage. The audit engine will alert you if the periods specified on the control group are insufficient to cover the forecast periods on the forecast definition. PeopleSoft Proprietary and Confidential 27

42 Setting Up Forecasting Chapter 4 Warning! Ensure that there are no overlapping values in the Forecast Value column for each row you create, unless you are creating a rule for a zero-growth forecast. To do this, specify 0 (zero) for the Forecast Value and Annual % Growth columns. Specifying the Exponential Smoothing Forecast Profile If you selected a Statistical Method of EXPSMOOTH on the Control Group page, access the Exponential Smoothing page. Exponential Smoothing page To specify the exponential smoothing forecast profile: 1. Select the Holt-Winter s Method: Additive, Double Exponential, or Multiplicative. 2. Select the Smooth Trend to use with this profile: Damped, Exponential, or Linear. 3. Enter the Alpha, Beta, Gamma,and Phi values, as appropriate, where Alpha is the level smoothing weight, Gamma is the trend smoothing weight, Beta is the seasonal smoothing weight, and Phi is the trend damping weight. 4. Specify the Season Length length of the cycle in when applicable. Using the Control Group Clone Utility You can also create control groups by copying an existing control group and making minor modifications. This is a great time saver if you are setting up control groups with similar options. You can copy control groups using the Control Group Clone Utility. 28 PeopleSoft Proprietary and Confidential

43 Chapter 4 Setting Up Forecasting Page Used to Clone Control Groups Page Name Object Name Navigation Usage Copy Control Group AF_COPYCG_PNL Analytic Forecasts, Analytic Forecasting Home, Set Up Forecast Rules, Control Group Clone Utility Copying a Control Group Create a control group by copying an existing control group with similar options. Access the Copy Control Group page by selecting the Control Group ID of the group you want to use to copy. To copy a control group: 1. Enter a New Control Group ID and New Description. 2. Click the Copy button. The system creates the new control group. You can now use the Define Control Group component to make the necessary modifications. Note. The new control group has the same effective date as that of the original. Defining Datasets As a next step in the process, you define the datasets on the Data Set Definition page. This page defines a dataset specifically for forecasting purposes, using metadata such as TableMaps, DataMaps, filters, and constraints defined in PeopleSoft Enterprise Warehouse. It is similar to the dataset definition page in Enterprise Warehouse, but contains fields necessary for the forecasting application. On the DataSet page you specify which history data you want to use to create a dataset, what dimensions to use to break the forecasts into individual elements, and what date field to use to measure time. See Also Page Used to Define Datasets Page Name Object Name Navigation Usage Data Set Definition AF_DATASET_DEFN Analytic Forecasts, Analytic Forecasting Home, Set Up Forecast Rules, Define DataSet Definition Setting Up Datasets Access the Data Set Definition page. Establish datasets for forecasting. PeopleSoft Proprietary and Confidential 29

44 Setting Up Forecasting Chapter 4 Data Set Definition page To set up datasets: 1. Select a Constraint Code. The constraint code prompts from the Constraint Definition and Constraint Criteria page. The page displays the fields available through the selected constraint. Note. Resolved fields (BUSINESS_UNIT, SETID, SCENARIO ID, MODEL_ID, ASOF_DT, and EFFDT) cannot be selected from this page. A warning will be issued if the record contains ASOF_DT, TRANSACTION_DT, or EFFDT since only one value of these keys will be available to the forecast in any job, because they are resolved through the run control. 2. For each Description field you want to use in the forecast, select its Select check box. For each check box you select, a drop-down list box displays in the Aggregate Type column. 3. Select the Aggregate Type you want performed for the item listed. The aggregate type determines how the forecasting program that imports history processes the field. Select from the following: Average Average Distinct The summary function that averages the numeric values in the group is applied. The summary function that averages in the group is applied. 30 PeopleSoft Proprietary and Confidential

45 Chapter 4 Setting Up Forecasting Count Count Distinct Group By Max(maximum) Min(minimum) Sum The summary function that counts the numeric values in the group is applied. The summary function that counts the number of unique values in the group is applied. Distinct values of the field are grouped together and the summary functions is applied. The summary function that returns the maximum value of the forecast quantity is applied. The summary function that returns the maximum value of the forecast quantity is applied. The summary function that sums the numeric values in the group is applied. Sum Distinct The summary function that sums the unique numeric values in the group is applied. 4. Click the Build Dataset SQL button to compile and integrate this dataset into the system. Note. Compilation and integration usually happen automatically when adding or editing a dataset. If nothing is edited, use the Build Dataset SQL button to force recompilation. Creating Forecast Definitions PeopleSoft Analytic Forecasting enables different analytic applications to use the same forecasting capabilities by using forecast definitions. One application will typically have just one forecast definition. When setting up forecast definitions, you select a calendar. For example, you can select that all forecasting be done on a monthly basis. A forecast definition operates in cycles similar to that of a budget. You also specify the forecasting method used to fit the history data to the model by selecting one or more control groups for use with that definition. This enables different products to be forecast using different methods. Here are the steps to set up forecast definitions using the Forecast Definition component: 1. Define the forecast definition parameters on the Definition page. 2. Establish the dataset code for the forecast on the Forecast Quantity page. 3. (Optional) Specify an independent variable for use in a multivariate forecast on the Independent Variables page. 4. (Optional) Specify the seasonality group to use on the Seasonality Group page. 5. Specify the control group for the forecast, which controls the statistical method to be used, on the Control Group page. 6. Schedule the forecast execution cycle on the ScheduleStatuspageandspecifyingany notes you want to include on the Notes page. Note. You can review the definitions you create by running the Forecast Definitions Crystal report. PeopleSoft Proprietary and Confidential 31

46 Setting Up Forecasting Chapter 4 See Also Appendix A, PeopleSoft Analytic Forecasting Reports, page 59 Pages Used to Create Forecast Definitions Page Name Object Name Navigation Usage AF_FCSTDEFN_PNL Analytic Forecasts, Analytic Forecasting Home, Set Up Forecast Rules, Define Forecast Rule Forecast Definition - Definition Forecast Quantity AF_FCSTDVAR_PNL Select the Forecast Quantity tab. Independent Variables AF_FCSTIVAR_PNL Select the Independent Variables tab. Seasonality Group AF_FCSTSG_PNL Select the Seasonality Group tab. Set up the specific criteria for your forecast definition, including calendar ID and forecast periods. Define from where in the dataset page this forecast will derive its values. Specify an independent variable for use in a multivariate forecast. Enter seasonal adjustment factors for the forecast definition. Control Group AF_FCSTCG_PNL Select the Control Group tab. Define at least one control group for the forecast definition. Schedule AF_FCSTSCH_PNL Select the Schedule tab. Schedule a forecast execution cycle. Notes AF_FCSTNOTE_PNL Select the Notes tab. Enter any setup comments, advice,ornotesforthe forecast definition. Forecast Definitions RUN_RPS_0001 Analytic Forecasts, Analytic Forecasting Home, Review Forecasting Reports, Forecast Definition Defining Your Forecast Definition Parameters Access the Forecast Definition - Definition page. Review forecast definitions you have established. 32 PeopleSoft Proprietary and Confidential

47 Chapter 4 Setting Up Forecasting Forecast Definition - Definition page To set up your forecast definition: 1. Enter a Forecast Status: Active Disabled The definition is available for use in forecasting. The definition is not currently available for use for forecasting. Setup Mode The definition can be modified arbitrarily, whereas once the definition is in use for forecasting, most values are fixed and cannot be changed. The Audit Forecast Definition process validates the forecast rules and, if the rules are all correct, changes the status from Setup Mode to Active. 2. Enter a Calendar ID from the prompted Calendar Builder page. Forecast values are calculated for each period in a year. 3. Select the Multivariate check box if you want to use at least one independent variable in regression forecast. The Independent Variables page becomes active enabling you to enter a variable. 4. Select the Imported check box if you are going to import the forecast with an ETL tool, instead of generating a forecast. 5. Select the Auto-create New Elements check box to create new elements automatically using the Import engine. If you do not select this check box, the system will not create new elements. If you want to import history without automatically creating elements, you must first explicitly create elements with keys that match the dimensions in your history on the Forecast Elements page. 6. Enter the maximum number of Total History Periods to be included in the calculation of the forecast values for the forecast definition. For example, if the value is 24, then the past 24 history periods will be used in calculating the forecast values. PeopleSoft Proprietary and Confidential 33

48 Setting Up Forecasting Chapter 4 Note. The number of history periods used for the forecast may not necessarily be the same as the number of history periods that are being imported, if history is in fact being imported. 7. Enter the number of Total Forecast Periods for which you want to in the future forecast values. If you are using a forecast-free model, specify 0 (zero) in this field. 8. Enter a Confidence Level % to control the size of the standard error. The confidence level should be between 51 and 99 percent. Confidence levels below 50 percent do not produce meaningful forecasts. 9. In the Forecast Order field, enter a number to specify the sequence in which forecast definitions should be processed. This parameter is useful in cases where one forecast provides input to another forecast. 10. In the Forecast Tracking Information group box, define the user responsible for resolving the exceptions list by selecting that user s ID in the Exceptions To field. Exceptions are conditions or data relationships encountered by a forecasting process that should be brought to the attention of the forecaster. See Also Establishing the Dataset Code Access the Forecast Quantity page. Forecast Quantity page To establish the dataset code: 1. Enter a DataSet Code. This prompt displays all the datasets you created on the Data Set Definition page. For example, if you are in the financial services industry, select an instrument; if you are using data for PeopleSoft Activity-Based Management purposes, select a product name. 2. Enter a Quantity to Forecast and an Assigned Date. The date specifies the date field used to select history data and to assign the data to fiscal years and accounting periods. 34 PeopleSoft Proprietary and Confidential

49 Chapter 4 Setting Up Forecasting 3. Specify the precision for the forecast by entering the number of decimal places for the forecasted values in the Decimal Positions field. 4. In the Type of Quantity group box, select the quantity type for the forecast. Once selected, the system associates the type with the forecast in order for other analytical applications to use it in their calculations. Percent #ofunits(number of units) Currency The forecast quantity is a percentage (for example, a rate). The forecast quantity is a number. The forecast quantity is measured in units of the standard currency. Not Applicable The forecast quantity has no relevant unit of measure. 5. If you are using PeopleSoft Analytic Forecasting with PeopleSoft Funds Transfer Pricing, Risk-Weighted Capital, or Asset Liability Management, select a Financial Interpretation type for the forecast. Once you select a type, the system associates the type with the forecast in order for other analytical applications to use it in their calculations. Originations - - Current Bal (originations current balance) Originations - - Commitment Bal (originations commitment balance) Originations - - Instr Count (originations instrument count) Not Applicable The quantity is the current balance of originations per fiscal year/accounting period. Commitment balance for instruments originated in the year/period. The number of instrument originations. No financial interpretation applies. Specifying an Independent Variable for a Multivariate Forecast Access the Independent Variables page. This page is only active if you select the Multivariate check box on the Definition page. To specify independent variables for a multivariate forecast: 1. In the Independent Variables group box select the Keyed check box to specify that the independent variable has the same keys, in the same sequence, as the dependent variable. 2. Specify a numeric value to indicate the number of finite Differences to the independent variable time series. 3. In the Lags column, specify the number of periods by which to delay the independent variable time series. 4. In the Variable Business Unit column, specify the business unit of the independent variable, and in thevariable Forecast Definition column, specify the forecast definition of the independent variable. 5. In the Days Stale column, specify the number of days to ensure that the forecast for the independent variable has been calculated within this many days of the dependent forecast. Specifying the Seasonality Group Access the Seasonality Group page. PeopleSoft Proprietary and Confidential 35

50 Setting Up Forecasting Chapter 4 To specify seasonality for the forecast definition: 1. In the Seasonality Groups group box, specify the seasonality group or groups you want to use for the forecast definition. You set up these groups on the Seasonality Group Definition page. 2. Select the Default check box to indicate that the seasonality group is the default for any new elements that are automatically calculated. If you do not select this check box, new elements that are created will not have seasonality applied to them. 3. Enter a Seasonality Group ID for this forecast definition. You can leave this column blank or you can select one or more seasonality groups for this forecast definition. 4. The Periods Per Cycle column displays the number of periods per seasonal cycle that apply to the selected seasonality group. Specifying the Control Group Access the Control Group page. 1. You must enter at least one row in the Control Groups group box. 2. For at least one of the rows you enter, select the Default check box to indicate that the control group is the default for any new elements that are automatically calculated. One control group must be designated as the default control group. Once a new element has been automatically created, you can assign a new control group for the element by using the Forecast Element setup pages. 3. Enter a Control Group ID for this forecast definition. This ID is prompted from the Control Group setup page. Scheduling the Forecast Execution Cycle and Entering Notes Access the Schedule page. To schedule a forecast execution cycle (a series of forecasting processes) for the forecast definition: 1. In the As Of Date column, assign an as of date to the forecast. This date must be in the active year and period. 2. In the Active Year column, enter the first year for which forecast values are calculated. 3. In theactive Period column, enter the first period for which forecast values are calculated. 4. The Cycle Status field shows the current status of this part of the cycle. The status is normally set automatically, but you can change the status if you change something and want to rerun a forecast. Completed Forecasting The cycle has completed. Computes seasonal variation and new forecast. Pre-forecasting Imports history and computes forecast accuracy. 5. In the Import History From Year column, enter the year from which you want your history data imported. 6. In the Import History From Period column, enter the period from which you want your history data imported. 7. Access the Notes page and enter any comments or notes for this forecast definition. Reviewing Your Forecast Definitions Access the Forecast Definitions page. 36 PeopleSoft Proprietary and Confidential

51 Chapter 4 Setting Up Forecasting To review any forecast definitions you have established, specify the SetID and Forecast Definition for the definition you want to review. Select the Use all Forecast Definitions? check box to report on all forecast definitions defined for the chosen SetID. If you select this, you do not need to specify any Forecast Definition. The reports lists the specifications for the forecast definition. Defining Forecast Groups Because the batch engines resolve data based on forecast groups, you must define at least one forecast group even if it contains only one forecast. To do this you use the Forecast Groups page, where you can also group together several forecast definitions for a scenario. You must set up a model ID for this forecast group, and a scenario, both in PeopleSoft Enterprise Warehouse. See Also Page Used to Define Forecast Groups Page Name Object Name Navigation Usage Forecast Group AF_FCSTGRP_PNL Analytic Forecasts, Analytic Forecasting Home, Set Up Forecast Rules, Define Forecast Group Setting Up a Forecast Group Access the Forecast Group page. Define forecast groups, and group together several forecast definitions for use in a scenario. PeopleSoft Proprietary and Confidential 37

52 Setting Up Forecasting Chapter 4 Forecast Group page To define a forecast group: 1. Specify a Calendar ID to apply to this forecast group. All forecasts definitions included in the group must use the same calendar. 2. Enter a Model ID to apply to this forecast group. All forecasts in the group must use the same model ID or one of the parent models. 3. The Forecasts in Group group box displays all available forecast definitions currently in the group. 4. The Available Forecasts group box displays the forecast definitions that can be added to the forecast group you are creating. 5. Use the left and right arrow buttons to move a single, selected forecast definition to either the Forecast in Group column or to the Available Forecasts column. 6. Use the All Left and All Right double-arrow buttons to move at one time all the forecast definitions to either the Forecast in Group column or to the Available Forecasts column. Using the Forecast Clone Utility You can also create forecast definitions by copying an existing definition and making modifications to it. This is a great time saver if you are setting up forecast definitions with similar options. You can copy forecast definitions using the Forecast Clone Utility. 38 PeopleSoft Proprietary and Confidential

53 Chapter 4 Setting Up Forecasting Page Used to Clone Forecasts Page Name Object Name Navigation Usage Copy Forecast Definition AF_COPYAF_PNL Analytic Forecasts, Analytic Forecasting Home, Set Up Forecast Rules, Forecast Clone Utility Copying a Forecast Definition Access the Copy Forecast Definition page by selecting the Forecast Definition ID of the definition you want to copy. Create a forecast definition by copying an existing definition with similar options. To copy a forecast definition, enter a New Forecast Definition, New Model ID, and New Descriptionand click the Copy button. The system creates the new forecast definition. You can now use the Forecast Definition component to make the necessary modifications. Note. The new forecast definition has the same effective date as the original. Defining Forecast Elements You can create forecast elements in one of two ways: automatically, by running the Import engine, or explicitly, using the Forecast Element component. Forecast elements are created automatically as they are found in the history data set. When elements are created automatically, PeopleSoft Analytic Forecasting assigns the default control group and, if applicable, the default seasonality group to the element. Later, however, you may want to change the defaults for a given element. After you complete the pre-forecasting phase, you can optionally define additional forecast elements. For example, you may want to create an element that has no history so that a forecast for it can be generated. The Forecast Elements setup component enables you to complete these requirements. Defining forecast elements involves: Creating key values. Defining performance parameters. Viewing the statistical, regression, or smoothing parameters, depending on the statistical method. Specifyingseasonality properties. See Also Chapter 5, Processing, Reviewing, and Adjusting Forecasts, Using Engines and Jobstreams, page 46 PeopleSoft Proprietary and Confidential 39

54 Setting Up Forecasting Chapter 4 Pages Used to Define Forecast Elements Page Name Object Name Navigation Usage View Keys AF_ELKEY_PNL Analytic Forecasts, Analytic Forecasting Home, Set Up Forecast Rules, Define Forecast Element Create or view key values for a forecast element. Performance AF_ELPERFM_PNL Select the Performance tab. Manage performance parameters for an element manually. Statistical Parameters AF_ELPARM_PNL Select the Statistical Parameters tab. Regression Parameters AF_ELREGR_PNL Select the Regression Parameters tab. Smoothing Parameters AF_ELSMTH_PNL Select the Smoothing Parameters tab. View statistical forecasting parameters for an element. View regressionparameters. View smoothing parameters. Seasonality AF_ELSG_PNL Select the Seasonality tab. Assign seasonality properties. Viewing and Creating Key Values Access the View Keys page. View Keys page To create or view key values for a forecast element: 1. Enter an Element Description to describe the element. For automatically created elements, the description should display all key values that comprise the element. 40 PeopleSoft Proprietary and Confidential

55 Chapter 4 Setting Up Forecasting 2. For automatically created elements, the Key Values fields are display only and reflect all the dimension values that constitute the unique element. For explicitly created elements, you must select Key Values that match existing dimensions in the Dimension tables. 3. The Available History Periods shows how many periods of history are available for forecasting. You can enter the number of History Periods To Use as well as the number of Forecast Periods To Use to forecast ahead. 4. Enter a typical or minimal forecast Base Value for use with the scheduled growth statistical method. 5. Specify the First Forecast Year, First Forecast Period, Last Forecast Year,and Last Forecast Period as applicable. The system will not produce a non-zero forecast for periods before the first forecast year, before the first forecast year and period, after the last forecast year, or last forecast year and period, respectively. Defining Performance Parameters Access the Performance page. The system will reset the element for a variety of reasons. For example, if you change the control group for the element, if you change the seasonality group for the element, or when the performance job has determined that the regression parameters are not accurate enough. Performance page To manage a forecast manually: 1. Select the Forecast Disabled check box to skip computing a forecast for this element. 2. Select the Forecast Reset check box to reset the forecast for the element. This means that when the forecast for this element is next calculated, the system will recalculate the statistical parameters from the history, instead of just extending the forecast based on the previous values of these parameters. If the Analytic Forecasting performance job determines that the regression parameters are not accurate enough, the system automatically selects the Forecast Reset check box. When the Forecast All (FSCTALL) or Forecast Reset (FCSTRST) jobs are run, the regression parameters (coefficients) will be recalculated (instead of carrying forward the original ones) in order to come up with forecast values. PeopleSoft Proprietary and Confidential 41

56 Setting Up Forecasting Chapter 4 Note. When you change the seasonality group for an element, the system will select the Forecast Reset check box automatically. If you do not want new regression parameters to be calculated, and you just want a new seasonality group to be used, then you must manually clear this check box and run FSCTALL to get a new forecast (FSCTSRST runs only those jobs that have the Forecast Reset check box selected). If you do not clear the check box, you will obtain all new parameters when you wanted only new seasonality factors. 3. Select the Forecast Always Reset check box to force the forecast to always be computed by recalculating the statistical parameters from history. 4. In the Threshold # Standard Errors field, enter a number for the maximum number of standard errors that will be tolerated without producing an exception when the forecast is compared to actual history. An exception will be created if the number of standard errors exceeds this amount. 5. The Max # Std Errs in Last AF Job (maximum number of standard errors in the last Analytic Forecasting job) field displays the number of standard errors in the forecast when last compared to actual history. Viewing Statistical Parameters Access the Statistical Parameters page. To view or specify statistical parameters for a forecast element: 1. Select the Control Group Sequence to use in forecasting this element. The sequence number reflects the row order on the Control Group page within a forecast definition. 2. The Statistical Method used by this control group is display only. 3. The Current Parameter Values group box displays the current values of the statistical parameters use to compute the forecast for Scheduled Growth forecasts. The parameter IDs and the meanings of their values are specific to the forecasting method. You can adjust the Parameter ID and Parameter Value for this forecast element. 4. The As of Date reflects the cycle data of the forecast run from Schedule Status page. This date applies only to scheduled growth forecasts. Multiple linear regression forecasts are accounted for on the Regression Parameters page. 5. You should rarely need to, but if you do need to change the value of a statistical parameter, click the Advanced button to make the parameter fields active. Make any necessary modification and save the page. Viewing Regression Parameters Access the Regression Parameters page. View the following fields: As Of Date ALPHA Sample Size Reflects the cycle date of the forecast run from Schedule Status page. This date applies only to multiple linear regression forecasts. A display only field that shows the ALPHA (intercept) parameter. Adisplay only field that shows the number of points used to compute the regression. 42 PeopleSoft Proprietary and Confidential

57 Chapter 4 Setting Up Forecasting ALPHA Error Degrees of Freedom Residuals Regression Coefficients A display only field that shows the adjusted standard error in the estimate of the ALPHA. A display only field that shows the number of degrees of freedom in the regression. A display only field that shows the sum of the squared residuals. This grid lists the Coefficient, Coefficient Standard Error, Variable Mean, and Covariances for each variable. Viewing Smoothing Parameters View the following data: Exponential Smoothing Coefficients This group box displays the exponential smoothing coefficient values and the initial seasonality estimation whenever they are applicable. The As Of Date reflects the cycle data of the forecast run from the Schedule Status page. Viewing Seasonality Factors Access the Seasonality page. To view or specify seasonality factors for the element: 1. Enter a Seasonality Group Sequence number. The sequence number reflects the row order on the Seasonality Group page within a forecast definition. The seasonality group displays in the Seasonality Group Used field. 2. Select the Seasonality Active check box if you want to use the history of this element in statistically computing the seasonal factors for the group. 3. Freeze Seasonality Factors is a display only check box. It is selected if you opted to not have seasonality factors recomputed with each forecasting cycle when you defined your seasonality factors. The system will continue to apply frozen default factors to the element in the seasonality group. 4. In the Seasonality Factors group box you can view the As of Date, Accounting Period, and associated Seasonality Factor for the seasonality group. PeopleSoft Proprietary and Confidential 43

58 Setting Up Forecasting Chapter 4 44 PeopleSoft Proprietary and Confidential

59 CHAPTER 5 Processing, Reviewing, and Adjusting Forecasts This chapter provides an overview of forecast creation and discusses how to: Use engines and jobstreams. Use tables for forecasting output. Use multicurrency functionality. Import historical data. Work with forecast processing. Adjust your forecast output. Reviewforecast data with Crystal reports. Overview Once your Analytic Forecasting setup is complete, processing takes place through PeopleSoft Enterprise Warehouse. This section assumes that you are already familiar with run controls, jobstreams, and the PeopleTools Process Scheduler. Review your PeopleSoft Enterprise Warehouse and PeopleTools documentation before proceeding if these terms are unfamiliar. Creating forecasts is an iterative process that comprises the following stages: Stage Description Pre-forecasting Includes all setup steps for compiling forecasts definitions, and the audit and import processes for the first forecast cycle. This includes the performance job for all subsequent cycles. If the import process completes cleanly, the performance process changes the cycle status from pre-forecasting to forecasting. Forecasting Includes recalculating and recomputing seasonality factors, all forecasts, and resetting the forecasts. PeopleSoft Proprietary and Confidential 45

60 Processing, Reviewing, and Adjusting Forecasts Chapter 5 Stage Description Adjustments Adjustments can be made to the history, forecasts, or seasonality. Adjustments to history and forecast values are pervasive across cycles; that is, if an adjustment is made in one cycle, it will carry forward to future cycles. Completion When you have a forecast that has been recomputed or adjusted, and finally meets your expectations, you can set the forecast definition schedule to completed. Nofurther processing will occur. Using Engines and Jobstreams PeopleSoft Analytic Forecasting uses several different Application Engine programs to process forecasts. This section lists the engines used, and the delivered jobstream definitions. These jobstreams are used to initiate processing from the PeopleSoft Enterprise Warehouse application. We also provide access to run control pages from the Forecast Processing home page. These pages enable you to run jobs directly from the PeopleSoft Analytic Forecasting application. See Also Engines Used in Analytic Forecasting PeopleSoft Enterprise Performance Management (EPM) applications use engines to run processes. We deliver engine metadata with your system. This metadata stores information about the various Application Engine programs used within an engine. The following table lists the delivered engines that are used by PeopleSoft Analytic Forecasting when processing forecasts using the delivered jobstreams and their associated jobs. Engine ID Name Usage AFAUDIT Audit Rules engine Audits the data setup rules for accuracy. AFIMPORT Import engine Imports history data from external sources. AFPERFM Forecast Accuracy engine Creates forecasts and checks the accuracy of the forecasts. AFSEASON Seasonal Factors engine Calculates seasonal factors for the forecast. 46 PeopleSoft Proprietary and Confidential

61 Chapter 5 Processing, Reviewing, and Adjusting Forecasts Engine ID Name Usage AFFCSTALL Forecast Group engine Calculates the forecasts within all forecast groups. AFFCSTRST Reset Forecasts engine Calculates reset forecasts. Using Jobstreams PeopleSoft Analytic Forecasting uses jobstreams for processing. Jobstreams enable different users to run a series of engines at the same time by using temporary tables. Temporary tables, known as record suites, use only the respective relevant data from the main Enterprise Warehouse fact tables to run the engines and process the data. This enables the engines in the jobstream to run faster while keeping the fact tables open and accessible to other users so they can run the same engines simultaneously. The system assigns the record suites when the first engine runs, and releases them when the last engine finishes. You associate each main engine with a job ID on the Job Metadata page. You also associate jobs with a jobstream on the Jobstream setup page. A job must be unique across all jobstreams. The following table lists each batch process, the delivered jobstream, and the engines and job IDs used in each jobstream. Process Delivered Jobstream ID Job ID and Engine ID Audit Forecast PF_JOBSTREAM AFAUDIT Compute All Forecasts PF_JOBSTREAM AFFCSTALL Recompute Reset Forecasts PF_JOBSTREAM AFFCSTRST Import History for Forecast PF_JOBSTREAM AFIMPORT Assess Forecast Accuracy PF_JOBSTREAM AFPERFM Update Seasonality Factors PF_JOBSTREAM AFSEASON Using Analytic Forecasting Run Control Pages In addition to using the jobstreams, you can run your Analytic Forecasting engines directly from the Forecast Processing main page by accessing the generic Enterprise Warehouse run control page (RUN_PF_JOBSTREAM). The following table lists the pages, respective job IDs, and description of each. PeopleSoft Proprietary and Confidential 47

62 Processing, Reviewing, and Adjusting Forecasts Chapter 5 Run Control Page Job ID Description Rule Validation Audit AFAUDIT Validates your forecast rule setup. Import Historical Forecasts AFIMPORT Imports historical data for forecasts. Forecast Accuracy Assessment AFPERFM Assesses the accuracy of your forecast. Seasonality Factor Update AFSEASON Updates forecast seasonality factors. Forecast Calculation AFFCSTALL Computes forecast values. Calculate Reset Forecasts AFFCSTRST Recomputes reset forecasts. Running Jobs from the Run Control Pages To run a job from the run control pages: 1. Select Analytic Forecasts, Analytic Forecasting Home, Forecast Processing, and then click the appropriate link. 2. Indicate When you want to run this job: Always, Don t, or Once. Usually you select Always. 3. When running Analytic Forecasting engines, do not select the As Of Dated Jobstream check box. Instead, enter the Fiscal Year and Period fields to process only those forecasts that have a schedule with that active year and period. 4. Provide a Description and specify the business Unit and Scenario ID to process. 5. Select the Jobstream ID to use. The jobstream IDs are provided in the table in the preceding section. 6. Select the Rerun check box, if you are processing the same job an additional time using the exact same parameters and want the system to re-resolve the tables. This option may slow down processing if you are assigned the same record suite. 7. Click the Run button to submit the job to the Process Scheduler. 8. When the Process Scheduler Request page appears, set the page controls for the process and click the OK button. Reviewing Forecast Jobstream Messages You can review messages generated during a job directly from the Forecast Processing main page by clicking the Forecast Jobstream Messages link. When you select the process instance you want to review, the system transfers you directly to the PeopleSoft Enterprise Warehouse engine messages component. This component is described in detail in your Enterprise Warehouse documentation. 48 PeopleSoft Proprietary and Confidential

63 Chapter 5 Processing, Reviewing, and Adjusting Forecasts See Also Tables Used for Forecasting Output Forecasting output from the following engines is stored in the following tables: Engine Table Contents AFIMPORT Element (AF_ELEMENT_F00) The elements in the forecast definition. AFFCSTALL and AFFCSTRST Analytic Forecasting (AF_FCST_F00) The actual forecast values and standard errors. AFFCSTALL and AFFCSTRST Analytic Forecasting Regression (AF_REGRSN_F00 and AF_REGRCOEF_R00) The regression statistics for use in simulation scenarios for regression testing. AFIMPORT Analytic Forecasting Dimension (AF_DIM_TBL) The dimension names that map to the dimension data stored on the element. Using Multicurrency Functionality PeopleSoft Analytic Forecasting can handle multiple currencies when creating forecasts. For example, if you want to set up a forecast in one currency, but ultimately want to report in a different currency, you can run the Currency Conversion engine to ensure that your forecasts will reflect the accurate currency. To use multicurrency functionality: 1. Create a forecast dataset to use a currency code (transaction currency) constraint with an aggregate type of Group By. 2. Run the forecast cycle. It will use the transaction currency in its calculations. 3. If you have other PeopleSoft analytic applications for the financial services industry (such as PeopleSoft Funds Transfer Pricing, Risk-Weighted Capital, or Asset Liability Management), you should run the Forecast Attribution engine to use the PeopleSoft Analytic Forecasting results to populate forecasted product originations, with both the Transaction field and the Base Currency field represented in the transaction currency. 4. Run the Currency Conversion engine. This will replace the base currency amounts in the pool tables with the correct base amount, according to your currency conversion rules. 5. Run the PeopleSoft Funds Transfer Pricing or PeopleSoft Risk-Weighted Capital engines on the forecasted product pools. PeopleSoft Proprietary and Confidential 49

64 Processing, Reviewing, and Adjusting Forecasts Chapter 5 Note. Because datasets are user-defined, you must set up your forecast dataset with Group By for the currency code, if you have business units that operate in currencies other than your base currency. See Also Chapter 4, Setting Up Forecasting, Defining Datasets, page 29 Importing Historical Data PeopleSoft Analytic Forecasting enables you to import forecasts from external sources such as a Microsoft Excel spreadsheet. In turn, this data becomes history data used in the generation of forecasts in Analytic Forecasting. To do this, use your PeopleSoft Enterprise Warehouse ETL tool to load the ODS tables and then run Data Loader to populate the data warehouse table for the dataset representing the forecasted data source. Once that dataset has been loaded with the externally forecasted data, use the Import History engine (AFIMPORT)to populate the Analytic Forecasting and Element tables (AF_FCST_F00 and AF_ELEMENT_F00) for the imported forecast. To import forecasted data using an ETL tool: 1. Define the dataset into which this imported data will be loaded. 2. Define a forecast definition for this external forecast, and ensure that the Imported check box is selected on the forecast Definition page. 3. Include the external forecast definitions in a forecast group. 4. Set up any other necessary forecast information as you would for non-imported forecasts. 5. Create the ETL map from the external forecast source to the ODS tables in Enterprise Warehouse and import your forecasted data. 6. Run the Data Loader to populate the Enterprise Warehouse table used as the dataset source in your forecast definition. See Also Working with Forecast Processing All forecasting processes can be launched in a jobstream from the Forecast Processing page links in Analytic Forecasting, or the Run Jobstream process page in Enterprise Warehouse. You can also setup jobstreams with one or more jobs by navigating to the Jobstream page in Enterprise Warehouse. When processing forecasts you can: 1. Run the Audit Rules engine (AFAUDIT) to validate your system setup. When running this engine, use the Fiscal Year and Period fields to process only those forecasts that have a schedule with that active year and period, and a forecast status of Pre-forecasting. 50 PeopleSoft Proprietary and Confidential

65 Chapter 5 Processing, Reviewing, and Adjusting Forecasts 2. (Optional) Import historical data with the Import engine (AFIMPORT) for use in your forecasts. Note. Run the Audit Setup job before you import history data. 3. Run the Forecast Accuracy engine (AFPERFM) to create your forecasts and check the accuracy. 4. When running this engine, use the Fiscal Year and Period fields to process only those forecasts that have a schedule with that active year and period and a forecast status of Pre-forecasting or Forecasting. To assess the forecast accuracy of a cycle, forecast data is processed for Cycle 1 with year and period from Cycle 2. To accomplish this, you must enter a year and period on the run control page that pertain to the active year and period of the Cycle 1 period forecasted in time 5. Run the Seasonal Factors engine (AFSEASON) to calculate seasonal factors for your forecast. 6. After this process is run, the Freeze Seasonality Factor check box on the Seasonality Group page becomes irrelevant when calculating seasonality factors. For example, the first time you use the Seasonal Factors engine to adjust the seasonality and the system clears this check box, the calculation of the seasonality factor on the Seasonality Group page is based on history values of the active elements that are associated with the seasonality group. If you later decide to use the default factors you established on the Seasonality Group page to calculate a forecast, you can no longer clear the Freeze Seasonality Factor check box and readjust the seasonality. The only way to override this is to make the active elements that are associated with the seasonality group inactive on the Seasonality page within the Forecast Elements component and readjust the seasonality. Optionally, you can manually copy the default factors to the Seasonalities adjustment page from the seasonality group. 7. Run the Forecast engine (AFFCSTALL) to calculate the forecasts only. 8. Review your forecast output and adjust as necessary. 9. Run the Reset Forecasts engine (AFFCSTRST) to recompute reset forecasts only. Note. When you process subsequent forecast cycles with a different fiscal year and or accounting period, you must reset the forecast elements by selecting the Reset check box for the element and saving your changes before submitting the forecast process. See Also PeopleSoft 8.8 Enterprise Warehouse PeopleBook, Streamlining Processing with Jobstreams Viewing Charts and Printing Results The View Historical Charts and View Forecast Charts components both feature charts. To view these charts, the EPM charting tool must be installed. See PeopleSoft 8.8 Enterprise Warehouse PeopleBook, PeopleSoft Enterprise Warehouse Preface, Viewing Charts and Printing Results. PeopleSoft Proprietary and Confidential 51

66 Processing, Reviewing, and Adjusting Forecasts Chapter 5 Adjusting Your Forecast Output One of the most important features of PeopleSoft Analytic Forecasting is that it enables you to focus your attention on the elements that need adjustment or resetting. Sometimes, the computed forecast may not reflect known future events; for example, promotions or product sales. You can review the forecast for each element and make any adjustments that are necessary to obtain an accurate forecast. You can make adjustments to the history data going into the forecast, and you can make adjustments to the forecast once it is created. You can also make adjustments to seasonality factors, once they are calculated through the engine. Adjustments are for a fiscal period in a year and are applied to each successive forecast version. Adjusting the forecast is the third and last phase of the forecasting process. To adjust a forecast: 1. Review history data and forecast data and adjust if necessary. 2. View and update seasonality data if necessary. 3. View and process exceptions. Note. Adjustments to the history and the forecast consist of values that modify or replace whatever the computed forecast value may be. Adjustments made in previous cycles remain effective, unless they are removed. 52 PeopleSoft Proprietary and Confidential

67 Chapter 5 Processing, Reviewing, and Adjusting Forecasts Pages Used to Adjust Output Page Name Object Name Navigation Usage AF_HIST_NEW_PNL Analytic Forecasts, Analytic Forecasting Home, Forecast Results and Charting, View Historical Charts View and Adjust Historical Data Retrieve specific history data to review and adjust. From this page, you can access a separate adjustment page on which you can adjust the data. Adjust History AF_ADJHIST_PNL Click the Adjust History Adjust history data. button. History Adjustments AF_NOTES_SEC Click the Notes button. Add any notes associated with the history or adjustments you make. View and Adjust Forecast Data AF_FCST_NEW_PNL Analytic Forecasts, Analytic Forecasting Home, Forecast Results and Charting, View Forecast Charts Adjust Forecast AF_ADJFCST_PNL Click the Adjust Forecast button. Retrieve specific forecasts to review and adjust. From this page, you can access a separate adjustment page on which you can adjust the data. Adjust forecast data. Forecast Adjustments AF_FCST_NOTES_SEC Click the Notes button. Add any notes associated with the forecast or adjustments you make. Maintain Seasonality Factors AF_MTNSG_PNL Analytic Forecasts, Analytic Forecasting Home, Forecast Results and Charting, Update Seasonality Data Exceptions AF_EXCP_PNL Analytic Forecasts, Analytic Forecasting Home, Forecast Results and Charting, View Forecast Exceptions Adjust seasonality groups. View exceptions resulting from a process, and to mark as Worked those exceptions no longer requiring attention. Reviewing and Adjusting Forecast and Historical Data The View and Adjust Forecast Data page is virtually identical to the View and Adjust History Data page. Access the View and Adjust Forecast Data page. PeopleSoft Proprietary and Confidential 53

68 Processing, Reviewing, and Adjusting Forecasts Chapter 5 View and Adjust Forecast Data page To review and adjust your forecast or historical data: 1. Enter Selection Criteria for the data you want to review. For most forecasts, there are too many elements to view all at once. Use the Selection Criteria to select limited subset of elements for viewing. The fields are: From Year From Period To Year To Period As Of Date From Element To Element Starting year of the data. Starting period of the data. Ending year of the data. Ending period of the data. The run date of the forecast job. Starting forecast element of the data you want to adjust. Ending forecast element of the data you want to adjust. Control Group Sequence number of the control group you established on the Control Group page. 2. Once you enter information in the required fields, click the Run Query button. This button disappears after you click it, and the system renders a forecast chart graphic. 54 PeopleSoft Proprietary and Confidential

69 Chapter 5 Processing, Reviewing, and Adjusting Forecasts 3. To render a chart of another element and change the grid data, scroll to the next element using the arrows on the Forecast Data Selected group box header. 4. To refresh the chart display, click the Refresh Chart button. Note. Scrolling through elements does not automatically refresh the chart display. 5. If you want to make any adjustments to the data, click the Adjust Forecast button to access the Adjust Forecast page. 6. On the Adjust Forecast page, enter the first year and period to which the adjustment applies in the Year and Period fields. 7. Enter the Number Periods (number of periods) for which the adjustment applies, and the Adjustment Type: Add Multiply Add a constant to the value. Enter a negative number to subtract. Multiply the value by a constant factor. Replace Replace the value by a different value. 8. Enter the Adjustment Amount. The User ID field is display only and shows the user ID of the individual making the adjustment. 9. Click the Notes button to access the Forecast Adjustments page on which you can enter a description of the adjustment. Viewing and Updating Seasonality Data Access Maintain Seasonality Factors page. Maintain Seasonality Factors page To view and update seasonality data: PeopleSoft Proprietary and Confidential 55