Data Mining: Level I

COURSE OUTLINE
SESSION I - Strategic Overview

This session introduces participants to the conceptual foundation of data mining projects. It is intended to give an overview of the types of problems that are appropriate for data mining, offer an approach that is realistic for applied model development in a business environment, and explain why traditional approaches are insufficient.

Why Build Models?

Data Mining: What it is, and What it isn t

Matching Technologies and Problem Types

Belief in Degree of Set Membership vs Right Answers

Why Traditional Statistics is Not Enough

Where Data Mining Works

Defining Goals for Better Performance


SESSION II It s About the Data

Improved performance generally comes from one of two sources: getting more information content from the available data, or getting better data. This session explores sources of data, the strengths and weaknesses of various types of data, and techniques for manipulating data to extract information content.

Types of Data

Sources of Data

Data Errors

Missing Data

Outliers

Normalizing Data

Derived Variables

Variable Reduction

Data Transformation Ideas


SESSION III Conceptual Introduction to Core Modeling Technologies

The third session introduces a number of the advanced technologies commonly used in data mining. Discussion of these technologies in Level I is focused on conceptual understanding, and the strengths and weaknesses of the various methods. A technical drill-down into techniques for the various algorithms is reserved for the Level II offering.

Linear Regression

Logistic Regression

Clustering

Classification Trees

Chaos

Neural Networks

Genetic Algorithms


SESSION IV Making Predictive Analytics Work

The final session addresses the Experimental Design and Project Definition aspects of Data Mining projects. These areas are where most data mining projects fall short of their potential.

Experimental Design

The Data Mining Process

Case Studies