HORIZON Business Intelligence The Art of Analysis Ryan Hodges, HORIZON 360 BI Development Manager 5/22/2017
Objective Apply the tools of an artist to the work of a data analyst Art Color Form Flow Narrative Analysis Breakdown Pattern-finding Root Causes Relationships 2
Analyzing With Your Whole Brain Computers have: CPU for math-based operations GPU for graphical processing Brains have: Frontal lobe for executive functions Parietal lobe for math-based operations Visual cortex for vision 3
More Than Pretty Pictures Visualizations Tell a story Focus attention Make an impression Make users think and feel SMART 4
Objection 1: My Data Is Too Messy and Complex Visualizations help make sense of messy data. Even particle physics relies on them. Gravity Waves Higgs Boson Predictions 5
Objection 2: I Don t Have Time For All This Visual Stuff Visual analysis is easy and time-saving 6
Objection 3: Pictures Are For Users, Not Experts A-ha! moments are for analysts too 7
COLOR 8
Power of Color Color serves many purposes: Effectively encodes data Triggers mental associations Makes reports more interesting Divides a country (no, not really) 9
The I Spy Game How long does it take you to spy the TOP/BOTTOM officers by Past Due % in this report? Beware data dumping It has been known to cause: Attention fatigue Answer blindness Indecisiveness Irritable Manager Syndrome 10
Color Vision Past Due Report revisited Pop-out effect Gets your attention Nowhere to hide in the data 11
Conditional Formatting Easy to use Preset formats Flexible Try it and see how quick and easy it is. 12
Choosing the Right Colors Use high contrast colors Simple palette Follow style guides Tap into the visual lexicon Red light / green light Arrows Icons 13
Visual Indicators Symbols provide: Context Target tracking Status Feedback 14
Color Banding User color bands for: User driven discovery Busy intersections (crosstabs) Making sense of big data sets Filtering out the irrelevant 15
Color Banding in Practice Comparing Customer Age Groups for Deposits and Loans 16
FORM 17
Basic Chart Food Groups Pies Composition Bars Comparison (side-by-side) Composition (stacked) Trends (series) Lines Trends Composition (area) 18
Pie Charts: Parts-to-whole Easy to build 1 attribute 1 measure Delivers visual punch Expand with Series Radial Add flexibility with Series Parameters Drillthroughs 19
Pie Charts: Mortgage Analysis Example Breaking it down: Swap out fields for on-the-fly analysis Show multiple perspectives 20
Bar Charts: Comparing Categories Works great with list reports 1 measure 1 or 2 attributes Supports groups and hierarchies Supports ordering Variations Vertical / horizontal Stacked Combination with line chart 21
Stacked Bar Charts Stacked Bars Categories Parts-to-whole Progression 22
Creating Your Own Categories Create with Custom groupings Calculated fields Combine multiple elements Lump smaller categories together Special case handling 23
Line Charts: Trending Ideal for time series Head-to-head comparisons Variations Area Stacked Area Combo with Bar 24
Bubble Charts: Relationships Works well with crosstabs 1-2 categories 1 size measure 2 attributes Show correlations Reveal outliers 25
WordCloud Great for label data such as: Officer Branch Product Coming soon! 26
Maps Market opportunities Coverage areas Hot spots Coming soon! 27
Advanced Charts Many advanced and special purpose charts are available: Heatmap Gauge Bullet Radar Decision Tree And many more 28
Choosing the Right Visual Representation Start by asking: What do I want to find out? Which visualization is the closest match? Who is my audience? Refine by asking: What did the data reveal? Have I analyzed it from all angles? 29
DATA 30
Data Selection Choose your data carefully: Quality Specificity Applicability Redundancy 31
Assessing your Data HIGH Usability LOW Usability # of Categories Distribution Missing Values Outliers 32
Rapid Prototyping Use ad hoc tools to: Explore data Create mock-ups Identify problems 33
Don t Forget to Verify Create balancing reports/queries Never stray too far from home-base Test often! 34
Ratios and Summaries Contextualize Divide & Conquer Set significance thresholds 35
Working with Rollups Tame large categories Enable multi-level analysis Work with data swathes 36
Working with Hierarchies General Ledger Structure Nested rollups Parent-child relationships Chart of Accounts Business Units 37
User Prompts Add extra data elements without: Taking up space Adding complexity Move between peer groups Facilitate self discovery 38
Advanced Prompts Did you know prompts can: Create what-if scenarios Be used in data calculations Apply weights Provide raw user inputs 39
FLOW 40
Break Up Reports Summary Detail 41
Seeing the Forest and the Trees Use drillthroughs to step into detail level data 42
Give Your Reports a Visual Makeover Identify reports that contain graphic ready elements: Ratios Sections Categories Summaries Time Series 43
Creating Clickpaths Drillthroughs can chained to: Walk a rollup or hierarchy Drill multiple levels Follow a train of thought Make a lateral pass 44
Interactions Reports and widgets can interact through: Shared prompts Common data elements Drillthroughs Coming soon! 45
Creating Workflows Create a Portfolio Build a library of reusable widgets Build personal dashboards Define custom tabs 46
Development Update 47
2016 Development in Review User Identifiable Information (UII) Encryption: SSN/TIN Driver s licenses and other IDs License audit automation New Data OAC (Online Account Creation) Interface Stockholder Info Multiple risk ratings 10 new data files 270 new fields 48
2017 Development Roadmap Coming this year: Cognos 11 GL Statistical Accounting CeB (Consumer ebanking) Integration 49
Coming Soon: Cognos 11 Arriving Q4 2017 One-time migration Fee No additional license or maintenance costs Submit an RPS (Request for Professional Services) to get on the implementation calendar 50
Cognos 11 Features Features HTML5 with full mobile support Single Studio concept Self guided analysis Revamped visualizations Improved search 51
Ryan Hodges ryan.hodges@fisglobal.com