Data Mining

Data mining, [also referred to as Statistics and Data Mining] refers to the interrogation of data for the purpose of identifying trends and patterns that indicate notable business activity. Statistical and data mining tools can perform predictive modeling or to d[iscover the cause-and-effect correlation between two metrics. This includes:

• Advanced Analysis
• Hypothesis Testing
• Predictive Analysis

Where relationships between data are known, data query is used. Data mining is employed when such relationships are not known.

Data Mining uncovers subtle relationships, such as price elasticity and sales trends using using set theory techniques, statistical treatments and other advanced mathematical functions. As an advanced BI tool, data mining is typically only performed by experienced analysts to perform:

• Correlation analysis
• Trend analysis
• Projections

The outcomes of data mining provide critical business insight for both strategic an tactical decision making. Data mining can be used across the business, for instance:

• Fraud detection
• Targeted marketing
• Risk management
• Business analysis and optimization

Example - Data Mining in Marketing

A marketing analyst may build a revenue model for a particular product group to show gross margins by quarter as a function of shipment times, pricing and demand. This can be used to estimate the financial impact of delayed shipments. The outcome will help the marketing team to determine any pricing changes required on current stock to help cover the cost of any shipment driven losses or determine promotional spend to push a substitute product until shipments can be normalized to a more profitable standard pattern.

Analysis or Data Mining tools are used to answer questions such as “What is the median spending of customers in each customer region?” or “What is our market share growth this year by store, for those stores
where actual sales exceeded target sales by more than 15%?”

Data mining tools learn about what is 'normal' in you business, and identify elements that do not conform with this pattern.

Data Mining Technology

The standard data mining interfaces are based on OLE DB for Data Mining specification and Data Mining Extensions [DMX] query language. These are widely used as standard interfaces to data mining objects and algorithms on various data mining platforms.

Data Mining Tool Constraints

Cube-based BI architectures - have inherent limitations that render them incapable of providing a comprehensive picture of the inter-relationships of data across the enterprise. This is especially so for massive data business models, such as those of Telecoms and Financial Services providers.

User interfaces - are fairly complex, however more recent tools are providing more usable interfaces. In spite of this, an understanding of statistical analysis is still required in most instances.

Data - The quality of the insights gained from Analysis or Data Mining tools is directly related to the quality and completeness of the underlying data. Calculations are performed by highly sophisticated SQL Generation Engines and a specialized Analytic Engine.

Microsoft Data Mining Tools

In the MS SQL Server 2000, Microsoft embedded data mining technology in its business intelligence platform. The more recent SQL Server 2005 platform uses the OLE DB and DMX standards to provide data mining capability.

SQL Server Analysis Services 2005 includes 9 algorithms for data mining:

1. Microsoft Association Rules
2. Microsoft Clustering
3. Microsoft Decision Trees
4. Microsoft Linear Regression
5. Microsoft Logistic Regression
6. Microsoft Naïve Bayes
7. Microsoft Neural Network
8. Microsoft Sequence Clustering
9. Microsoft Time Series

These algorithms can be applied to warehouse data stored SQL Server and other SQL platforms, or in multidimensional (OLAP) data stored in Analysis Services 2005. Since it can be queried directly with an open query language, the output from SQL Server data mining can be used in a variety ways.

Data mining info can be shown directly in reports, can be used to drive actions in other applications (such as suggesting cross-selling items in websites and Point-of-Sale systems), and can generate advanced data visualizations through interfaces in business analytical tools, SQL Server’s Business Intelligence Development Studio, and other graphical data interfaces.

NEXT: Advanced Analytics

Back To Top