TM Visual Data Mining with Mine Set Mine

TM Visual Data Mining with Mine. Set™ Mine. Set Web site URL: www. sgi. com/software/mineset 1

What is Mine. Set™? TM Visual data mining technology that helps your business quickly turn large amounts of data into actionable business insights Data Warehouses or Business Data Mine. Set™ Business Insights Visual Data Mining with Mine. Set™ 3/11/2021 2

Why is Mine. Set™ so important? TM Business Intelligence Solutions • Visualization is essential (IDC, 1998) – 80% of users find visualization to be desirable – 51% find it very or extremely desirable • Data mining algorithms (IDC, 1998) – Important to over 80% of data warehousing users • Explosive data growth (Meta Group, 1997) – Data warehouses double in size every 12 to 18 months • Scalability, CPU performance and I/O bandwidth (The Data Warehouse Institute, 1998) – Most important factors in selecting a data mining or data warehouse hardware platform Visual Data Mining with Mine. Set™ 3/11/2021 3

How does Mine. Set™ fit into Business Intelligence Solutions? TM Action/Feedback Layer Visualizations for Decision Makers Analytical Layer Visualization & Analytic Model Development for Business Analysts Descriptive Predictive Visual IT Infrastructure Layer ECTL Processes Operational System(s) Enterprise Data Warehouse Mart 1 Mart 2 Mart N Visual Data Mining with Mine. Set™ 3/11/2021 4

Data Mining Discovery Process TM Mine. Set Client GUI Selection Transformations Denormalized Data Subset Data Warehouse or Business Data Visual Data Mining Business Insights & Understanding Mine. Set Server Analytical Data Mining API to OLAP and Other Mining Algorithms Visual Data Mining with Mine. Set™ 3/11/2021 5

Data Mining Discovery Process with Mine. Set™ Mine. Set Clients • Windows • SGI IRIX Launch via TM Tool Manager (Controls, Visualizations) COM Active. X Mine. Set Servers • NT, Linux 32 -bit Single Threaded Data Transformations • SGI IRIX 64 -bit Parallel RDBMS Connections Oracle Sybase Informix ODBC Flat Files Data Warehouse & Business Data Analytical Data Mining Engine API to OLAP & Other Mining Algorithms Visual Data Mining with Mine. Set™ 3/11/2021 6

Mine. Set™ 3. 1 Key Features TM • Powerful Visual Data Mining – Visualizations launched from Mine. Set™ Clients or any Windows application or WEB browser • Insightful Analytic Data Mining – Classification, Regression, Association and Clustering data mining model development • Software Development Toolkit (SDK) – Plug-in APIs for Analytics, Visual, Transformations and Functions • Application Toolkit Extensions – APIs to facilitate the writing of Web enabled applications Visual Data Mining with Mine. Set™ 3/11/2021 7

Visual Data Mining with Mine. Set™ Statistics Visualizer Histogram Visualizer • Mean, Min/Max, Std. Dev. analysis • Distinct count & range analysis Visual Data Mining with Mine. Set™ 3/11/2021 TM 8

Visual Data Mining with Mine. Set™ Scatter Visualizer Splatter Visualizer • Multi-dimensional analysis for large data sets Visual Data Mining with Mine. Set™ 3/11/2021 TM 9

Visual Data Mining with Mine. Set™ Map Visualizer • Spatial trend analysis TM Tree Visualizer • Hierarchical trend analysis Visual Data Mining with Mine. Set™ 3/11/2021 10

Analytic Data Mining with Mine. Set™ TM Learning Supervised Mining (predictive & labeled columns) Classification Regression Discrete columns Continuous columns • Decision trees * • Regression trees • Evidence * • Option trees • Decision tables • Column Importance (column selection) Unsupervised Mining (descriptive & unlabeled columns) Association Clustering Correlations Segmentations • one-to-one • k-means • multi-way • iterative k-means * boosting option Visual Data Mining with Mine. Set™ 3/11/2021 11

Visualizing Analytic Data Mining Models with Mine. Set™ Decision Tree • Visualizer for Decision Tree data mining analysis TM Regression Tree • Visualizer for Regression Tree data mining analysis Visual Data Mining with Mine. Set™ 3/11/2021 12

Visualizing Analytic Data Mining Models with Mine. Set™ Evidence • Visualizer for Naïve. Bayes data mining results, interactive scoring & analysis TM Decision Table • Visualizer for Decision Table data mining analysis Visual Data Mining with Mine. Set™ 3/11/2021 13

Visualizing Analytic Data Mining Models with Mine. Set™ Cluster • Visualizer for Cluster data mining results analysis TM Association • Visualizer for LHS/RHS association i. e. , “market basket” analysis Visual Data Mining with Mine. Set™ 3/11/2021 14

Mine. Set™ Tool Manager -Integrating it all together TM Data Sources – Mine. Set Server & DB connections Data Destinations – Visualization Tools – Mining Tools – Data export Data Transformations – Remove, add, change or bin columns – Filter, Aggregate or sample columns – Apply Model or Plug-in Source 15 Visual Data Mining with Mine. Set™ 3/11/2021

Visualization Deployment Options with Mine. Set™ TM Visualizations can be launched via Mine. Set™ Clients from • Any Windows application through the COM protocol as Active. X components • Any Mine. Set™ client on Windows or SGI IRIX • WEB browsers through the COM protocol as Active. X components <CLICK HERE TO LAUNCH> • WEB browsers by recording the visualization in a web-media format or saving a snap shot Visual Data Mining with Mine. Set™ 3/11/2021 16

Data Mining Wave -Business Intelligence Solutions TM Data Warehousing Wave Data Mining Wave • Internally Focused • Reactive Problem Solving • Build data repositories to understand the past • Re-engineer infrastructures to support business operations and process workflow 1985 1990 1995 • Customer Focused • Proactive Solutions & New Market Development • Predict customer behavior, market trends, and competitive environments • Leverage IT infrastructure with Business Intelligence Solutions 2000 2005 Visual Data Mining with Mine. Set™ 3/11/2021 17

Close Loop Business Model Data Mining Pilot/Project Obtain Data Measure Results Mine. Set SGI PSO or Data Mining Partner Identify Business Indicators Investigate & Drill Down Implement Action Plan Approve Action Plan TM Create Action Plan Develop Visual/Analytical Models & Business Scenarios Visual Data Mining with Mine. Set™ 3/11/2021 18

Mine. Set™ Business Intelligence Solution Examples TM State of Texas Medicaid Fraud & Abuse Detection System • Situation Analysis: – About 25% of the Texas state budget goes to medical welfare programs – Estimated 10% of the $7. 3 B in Medicaid transactions are fraudulent – The previous system, Surveillance Utilization Revision Subsystem (SURS), detects only 14% of fraudulent providers • The Requirements – More accurate fraud detection and higher prosecution rates Visual Data Mining with Mine. Set™ 3/11/2021 19

State of Texas Welfare Fraud Management TM • The Business Intelligence Solution – ITC Fraud Spotlight data mining and fraud case management application – Mine. Set™ software for visualization – EDS System integration and management – SGI Origin 2000 high-performance IRIX server running Oracle 8 i, ITC & Mine. Set™ Server • The Results – Detection rate of suspected doubled to 38% – Solution paid for itself in less than 6 months – The fraud detection application being leveraged into other state programs Visual Data Mining with Mine. Set™ 3/11/2021 20

Clustering and Data Visualization with Mine. Set™ TM Data on 14, 000 providers analyzed by unsupervised neural networks Neural networks clustered providers based on 100+ columns Visualization tool displays clustering, showing known fraudulent providers Subset of 100 providers with similar patterns investigated: Hit rate > 70% Source: ITC 9 -22 -98 Visual Data Mining with Mine. Set™ 3/11/2021 21

State of Texas Welfare Fraud Management TM “These savings could help provide preventive care for several hundred thousand additional children” - Robin Herskowitz, Senior Policy Analyst with the Texas Office of the Comptroller of Public Accounts Visual Data Mining with Mine. Set™ 3/11/2021 22

Mine. Set™ Business Intelligence Solution Examples TM Bioinformatics scientists are using Mine. Set visual and analytical data mining to gain insights and understanding into genetic data. • Business Intelligence Solutions Examples – Analysis of Gene Expression Chip Data at Roche – Visualizing Gel Electrophoresis Data at the University of Michigan – Visualizing Sequence Comparisons at the EBI – Predicting Splice Junction Points Visual Data Mining with Mine. Set™ 3/11/2021 23

Analysis of Gene Expression Chip Data at Roche TM • Researchers at Roche use SGI Mine. Set to analyze and understand gene expression chip data. – Visualization: Gene expression chip data displayed using Mine. Set Scatter Visualizer. Image Courtesy Roche Group. Visual Data Mining with Mine. Set™ 3/11/2021 24

Visualizing Gel Electrophoresis Data at the University of MI TM • Professor Philip Andrews & Peter Ulintz, Biological Chemistry Dept. , are using the Mine. Set Splat Visualizer to view the large amount of electrophoresis data. – Visualization: Electrophoresis data displayed in 2 -dimensions using Mine. Set Scatter Visualizer. Image courtesy the University of Michigan. Visual Data Mining with Mine. Set™ 3/11/2021 25

Visualizing Sequence Comparisons at the EBI TM • The European Bioinformatics Institute uses Mine. Set to visualize the partial results of a segment-wise "allagainst-all" FASTA sequence comparison between two completed genomes. – Visualization: Genome comparisons using Mine. Set Map Visualizer. Image courtesy the EBI. Visual Data Mining with Mine. Set™ 3/11/2021 26

Predicting Splice Junction Points from the Gen. Bank Primate DB TM • Using splice junction information, the Mine. Set Decision Tree classifier is used to predict and identify splice junction points in other unknown primate sequences – Visualization: Evidence Visualization of DNA Splice junction data. Visual Data Mining with Mine. Set™ 3/11/2021 27

Predicting Splice Junction Points from the Gen. Bank Primate DB TM • Using splice junction information, the Mine. Set Evidence Classifier is used to predict and identify splice junction points in other unknown primate sequences – Visualization: Decision Tree Visualization of DNA Splice Junction Data. Visual Data Mining with Mine. Set™ 3/11/2021 28

Sample list of Business Intelligence Solutions using Mine. Set™ TM SGI Proprietary and Confidential Visual Data Mining with Mine. Set™ 3/11/2021 29

Mine. Set™ 3. 1 Summary TM • Powerful Visual Data Mining – Visualizations launched from Mine. Set™ Clients or any Windows application • Insightful Analytic Data Mining – Classification, Regression, Association and Clustering data mining model development • Scalable Client/Server Architecture – Windows and SGI IRIX Mine. Set™ clients – NT/Linux Mine. Set™ Servers (Single Threaded 32 bit) – IRIX Mine. Set™ Servers (Parallel 64 -bit) • Software Development Environment – APIs and plug-in interface and integration into other OLAP tools Visual Data Mining with Mine. Set™ 3/11/2021 30

Mine. Set™ 4. X Client Enhancements TM Future Client Enhancements • Visualization Environment – 2 D Visualizations • Data Mining Control Environment – Data Mining Project and Model management – Tool Manager Wizards for common Data Mining Tasks – Control API • Client API to provides access to all Mine. Set™ capabilities • Web enabling of server function • Enhanced configuration and launching of Data Visualization tools in WEB browser environments – Application Authoring Support Visual Data Mining with Mine. Set™ 3/11/2021 31

Mine. Set™ 4. X Server Enhancements TM Future Server Enhancements • Data Mining Analytics – New Analytics • Time Sequenced Data (e. g. , Association sequences) – Source Code export of Mine. Set™ Data Mining Models • Performance – Parallel Clustering – Parallel RDBMS, ODBC, Flat File, etc. to Mine. Set™ operations • Connectivity – Enhanced Direct RDBMS connectivity on NT, Linux and SGI IRIX Mine. Set™ Servers Visual Data Mining with Mine. Set™ 3/11/2021 32

Mine. Set™ Client/Server Platforms from sgi TM SGI Series 2000 SGI IRIX Servers SGI Series 1000 NT & LINUX Servers SGI Visual Data Mining with Mine. Set™ 3/11/2021 33

Visual Data Mining with Mine. Set™ 3/11/2021 34