THINKANALYTICS – Knowledge Discovery and Data Mining

 

Think EDM is a fully integrated Business Analysis Automation platform incorporating knowledge discovery and data mining techniques designed to provide business users with intelligent analysis capabilities. It combines ease-of-use with enterprise scalability to uncover latent knowledge from massive amounts of data through host applications. The advanced statistical, data mining and visualization techniques, coupled with the data extraction, transformation, integration, analytical and action components of Think EDM will allow Thornberry Consulting to provide a unique solution for any complex task.

Think EDM is a component-based platform for the development of data mining and predictive analytic solutions that can be embedded in business applications and deploy predictive intelligence to a large number of business users. An extensive library of components is provided to support the complete knowledge discovery process, from data selection, transformation and processing through to data mining and visualization. The component architecture enables users to add specialized components. Think EDM has the ability to process vast amounts of data and delivers real-time responses, eliminating performance problems that plague traditional data mining products.

The example below illustrates how customer data from multiple business sources shown on the left-hand-side (Mortgage, Investments, General Insurance and Life Insurance groups) is merged using ThinkAnalytics data connectors and written to the data warehouse. Alternatively, the merged data can also used directly to create a product association model (also shown below).

The Think EDM component architecture provides unlimited flexibility.  It transforms and cleanses data and deploys advanced analytical and visualization techniques, and provides the ability for users to add specialized components with ease.  The component library contains over 200 individual components, grouped with respect to their function, including:  

·        Data Access

·        Data Reduction

·        Data Transformation

·        Expressions

·        Date and String Handling

·        Algorithms (Associations Rules, Classification, Clustering, Forecasting, Decision Tree, various Neural Networks)

·        Statistics (Simulation Sampling, Regression, Statistical Tests, Summary Statistics, Correlation/Association)

·        Basket Analysis

·        Visualization

·        Connectors

·        Meta-Data

·        XML Transformations

Think EDM is an open knowledge discovery and data mining platform that utilizes industry standards such as PMML, HTML, XML and ODBC/JDBC. The system is cross platform and targets multiple business channels. The Think EDM platform provides an Application Programming Interface (API) to the complete environment, including the component library.  There are no data access restrictions.  The components provide access to a range of data formats, including the leading RDBMS platforms – DB2, Oracle, SQL Server and Teradata.

Listed below are some examples of the analytical and statistical functionality included in Think EDM:

·         Support for Random, Systematic, Cluster and Stratified sampling

·         Addition of Simple, Linear and Holt-Winters Exponential Smoothing forecasting

·         Integration with statistical product 'R'

·         Addition of One and Two Sample Kolmogorov-Smirnov (K-S) Test, One Way ANOVA and ChiSquare Independence Test

·         Added support for ChiSquare, F, Lognormal, Normal, StudentT, Exponential, Weibull, Pareto, and Poisson distributions

·         Extensive support for probability distribution data generation, sampling, estimation lookup and comparison.

·         Added support for financial risk analysis using Monte Carlo simulation to study confidence limits in distribution analysis.

 

Components form the basic building blocks for defining a knowledge discovery process. Each component performs a specific task, such as data reduction, data mining or visualization, and the Think EDM platform provides the ability to link a series of components together to perform a complete analytical process. 


Below are some examples of Visualization Windows associated with specific Think EDM components:

 

 

 

 

 

Knowledge discovery and data mining techniques are used to identify and exploit useful patterns in massive data volumes. Think EDM allows users to significantly increase the value of data by identifying complex non-linear patterns from a data warehouse, CRM front-office, eCommerce and other corporate systems. Delivering business knowledge enables users to leverage advanced data mining techniques through their business application without data mining experience. Delivering intelligent information for front-, back- or web-office requirements are essential in today’s ultra competitive business environment.

 

This product provides the key to unlocking the knowledge hidden deep within corporate systems and provides a seamless integration into traditional and e-business systems. The distributed nature of Think EDM can fully utilize all available resources and distributes required processes which makes it unique in today's current data mining offerings, and provides Thornberry Consulting with a powerful analytic tool to provide the required analytic reports.