Deployment. 15 Feb Data & Intelligence Global One Team. NTT DATA Mathematical Systems, Inc. NTT DATA Mathematical Systems, Inc.

Transcription

1 Data & Intelligence Global One Team NTT DATA Mathematical Systems, Inc. Deployment 15 Feb NTT DATA Mathematical Systems, Inc.

2 NTT DATA Mathematical Systems Inc. offers you 1. Optimization Let s PoC 2. Analytics Self-service 3. Deployment Platform to appear 2

3 Deployment : NTT Data Mathematical Systems, Inc. 1.Our History 2.VAP s Concepts Platform for Mathematical Solutions Cycle of Trial-and-Error Phase and Deployment Phase 3.VAP s Future New VAP is Coming 3

4 Our History 4

5 Visual Analytics Platform is a Platform for Mathematical Solutions Began to be developed in the late 1990s Initially born in 2002 as general data mining tool Needed by professionals in various trades as powerful Visual Programming Environment 5

6 Their Needs and Desires Become Increasingly Complicated Providing functions for Mathematical Solutions as Combinations Data Mining Machine Learning Optimization Simulation Capable of Rapidly Deploying and Using the solutions made Reborn in 2011 as integrated Platform for Mathematical Solutions 6

7 VAP s Concepts 7

8 Powerful Platform for Mathematical Solutions These products work on the Visual Analytics Platform and can connect with each other. Analysis Optimization Simulation Text Mining Studio Patent Mining express Visual Mining Studio Big Data Module BayoLinkS Visual R Platform Deep Learner Nuorium Optimizer FIOPT S-Quattro Simulation System Credit NASA Platform for Mathematics + Computer Science 8

9 VAP has Great Advantages You can use our products seamlessly by putting product icons on the VAP project board and connecting them with arrows. VAP Web Server You can make web applications from your analytics flow on VAP with ease. 9

10 PDCA Cycle of Data Analytics The application of analytics in business is one of the most valuable! Collecting and redoing analysis of feedbacks are also just as valuable! Plan Make & Change analytics flow Action Finding problems Do Deployment System Implementation PDCA cycle of data analytics is a key point of making values by using the analytics system. Check Check effectivity 10

11 In the case of Customer Scoring I would like to model customer scoring based on POS data I m contacting customers in the order of their scores. Receive Orders! analyst I made it! contact center Not Receive Orders The performance is reasonably good. Is it OK and Completed? NO IT S NOT We have to re-model and refine based on feedback because of The Models Tendency To Become Rapidly Outdated! 11

12 Cycle of Trial-and-Error Phase and Deployment Phase Analytics (Trial-and-Error) Phase Deployment Phase What users do What users do Analytics design Changing data, type, format, etc Model selection Parameter Tuning View selection With Ease Routine execution Frequent execution Fixed parameter Fixed View Fixed format What they need What they need Making analytics flow Flexible Partial execution Many models and algorithms Flexibility of parameter tuning Many graphs With Feedbacks Simple UI Intelligible UI Clear result view Parameter Tuning UI (Just a little) Cooperating with other systems 12

13 VAP s Future 13

14 New VAP is Coming! Parallel Execution and Multi Processing Unlimited executions per user Multi processing is possible in execution (depending on PC) Multi Layers Various combinations in your needs REST API/Socket.IO/Python API Multi Platform Frontend implemented by HTML5 Backend implemented by Node.js and Python (Excepting analytic algorithms) Windows/Linux/Mac OS (2021 Later) Multi Language Japanese/English/Chinese (2021 Later, or 2020 depending YOUR NEEDS) 14

15 New VAP Architecture Browser Well designed GUI for domain specific modeling GUI For Data Pre-Processing GUI For Analysis GUI For Optimization GUI For Simulation GUI For Other Application Other Application REST API WEB Server Socket. IO Selectable interconnections as your system needs Client Side Server Side Task Management Layer Python API TMS VMS BDM BayoLink VRP Deep Learner NuOPT S4 Others... OS/HW Linux Windows Mac OS Others on-premises AWS Other Clouds 15

16 Operation of New VAP Now New VAP Stand Alone Edition Stand Alone Edition Client-Server Edition Innovate! Internal connection Client-Server Edition Platform Server Network Network Remote desktop Platform Server 16

17 Combinations of New VAP What is a Characteristic of Requests Your System Needs? Characteristic High Load and Low Frequency Low Load and High Frequency Embedding in Edge etc. Solution REST API Socket.IO Python API New VAP offers Best Solutions for you needs! 17

18 REST API and Socket.IO in Use Cases Business System REST API Edge - Calling scheduled tasks, such as daily tasks Network - Offering GUI Socket.IO Platform Server Calling Asynchronously and Bidirectionally WITHOUT CHANGES you can use solutions you have made whenever! 18

19 Python API in a Use Case Sensors and Controllers Edge Standalone operation or Rapid response needed from msi.common.streaming import Server from msi.common.dataframe import DataFrame from msi.vms.modeling import load_model class PredictServer(Server): def init (self, model, port): super(). init (self, port) self. model = model def message(self, request, response): record = DataFrame(request.message) model = self. model fitted = model.predict(record) response.json(fitted) Call directly dtree = load_model( pretrained_dtree_model ) server = PredictServer(dtree, 40000) server.serve_forever() Python API VMS BDM Deep Learner TMS Others... VRP BayoLink Processing sensor streams and Controlling based on analyzed feedback NuOPT S4 19

20 New Visual Analytics Platform is Coming Soon. Experience accelerating your business on cycles of trial-and-error and deployment with New VAP. Don t miss it! 20

21