Data Mining Techniques -- Contents -- Introduction -- Chapter 1 What Is Data Mining and Why Do It? -- What Is Data Mining? -- Data Mining Is a Business Process -- Large Amounts of Data -- Meaningful Patterns and Rules -- Data Mining and Customer Relationship Management -- Why Now? -- Data Is Being Produced -- Data Is Being Warehoused -- Computing Power Is Affordable -- Interest in Customer Relationship Management Is Strong -- Every Business Is a Service Business -- Information Is a Product -- Commercial Data Mining Software Products Have Become Available -- Skills for the Data Miner -- The Virtuous Cycle of Data Mining -- A Case Study in Business Data Mining -- Identifying BofA's Business Challenge -- Applying Data Mining -- Acting on the Results -- Measuring the Effects of Data Mining -- Steps of the Virtuous Cycle -- Identify Business Opportunities -- Transform Data into Information -- Act on the Information -- Measure the Results -- Data Mining in the Context of the Virtuous Cycle -- Lessons Learned -- Chapter 2 Data Mining Applications in Marketing and Customer Relationship Management -- Two Customer Lifecycles -- The Customer's Lifecycle -- The Customer Lifecycle -- Subscription Relationships versus Event-Based Relationships -- Event-Based Relationships -- Subscription-Based Relationships -- Organize Business Processes Around the Customer Lifecycle -- Customer Acquisition -- Who Are the Prospects? -- When Is a Customer Acquired? -- What Is the Role of Data Mining? -- Customer Activation -- Customer Relationship Management -- Winback -- Data Mining Applications for Customer Acquisition -- Identifying Good Prospects -- Choosing a Communication Channel -- Picking Appropriate Messages -- A Data Mining Example: Choosing the Right Place to Advertise -- Who Fits the Profile? -- Measuring Fitness for Groups of Readers.

Data Mining to Improve Direct Marketing Campaigns -- Response Modeling -- Optimizing Response for a Fixed Budget -- Optimizing Campaign Profitability -- Reaching the People Most Influenced by the Message -- Using Current Customers to Learn About Prospects -- Start Tracking Customers Before They Become "Customers" -- Gather Information from New Customers -- Acquisition-Time Variables Can Predict Future Outcomes -- Data Mining Applications for Customer Relationship Management -- Matching Campaigns to Customers -- Reducing Exposure to Credit Risk -- Predicting Who Will Default -- Improving Collections -- Determining Customer Value -- Cross-selling, Up-selling, and Making Recommendations -- Finding the Right Time for an Offer -- Making Recommendations -- Retention -- Recognizing Attrition -- Why Attrition Matters -- Different Kinds of Attrition -- Different Kinds of Attrition Model -- Predicting Who Will Leave -- Predicting How Long Customers Will Stay -- Beyond the Customer Lifecycle -- Lessons Learned -- Chapter 3 The Data Mining Process -- What Can Go Wrong? -- Learning Things That Aren't True -- Patterns May Not Represent Any Underlying Rule -- The Model Set May Not Reflect the Relevant Population -- Data May Be at the Wrong Level of Detail -- Learning Things That Are True, but Not Useful -- Learning Things That Are Already Known (or Should Be Known) -- Learning Things That Can't Be Used -- Data Mining Styles -- Hypothesis Testing -- Generating Hypotheses -- Testing Hypotheses Using Existing Data -- Hypothesis Testing and Experimentation -- Case Study in Hypothesis Testing: Measuring the Wrong Thing -- Directed Data Mining -- Undirected Data Mining -- Goals, Tasks, and Techniques -- Data Mining Business Goals -- Data Mining Tasks -- Preparing Data for Mining -- Exploratory Data Analysis -- Binary Response Modeling (Binary Classification).

Classification -- Estimation -- Finding Clusters, Associations, and Affinity Groups -- Applying a Model to New Data -- Data Mining Techniques -- Formulating Data Mining Problems: From Goals to Tasks to Techniques -- Choosing the Best Places to Advertise -- Determining the Best Product to Offer a Customer -- Finding the Best Locations for Branches or Stores -- Segmenting Customers on Future Profitability -- Decreasing Exposure to Risk of Default -- Improving Customer Retention -- Detecting Fraudulent Claims -- What Techniques for Which Tasks? -- Is There a Target or Targets? -- What Is the Target Data Like? -- What Is the Input Data Like? -- How Important Is Ease of Use? -- How Important Is Model Explicability? -- Lessons Learned -- Chapter 4 Statistics 101: What You Should Know About Data -- Occam's Razor -- Skepticism and Simpson's Paradox -- The Null Hypothesis -- P-Values -- Looking At and Measuring Data -- Categorical Values -- Histograms -- Time Series -- Standardized Values (Z-Scores) -- From Z-Scores to Probabilities -- Cross-Tabulations -- Numeric Variables -- Statistical Measures for Continuous Variables -- Interesting Properties of the Average and Median -- Variance and Standard Deviation -- A Couple More Statistical Ideas -- Measuring Response -- Standard Error of a Proportion -- Comparing Results Using Confidence Bounds -- Comparing Results Using Difference of Proportions -- Size of Sample -- What the Confidence Interval Really Means -- Size of Test and Control for an Experiment -- Multiple Comparisons -- The Confidence Level with Multiple Comparisons -- Bonferroni's Correction -- Chi-Square Test -- Expected Values -- Chi-Square Value -- Comparison of Chi-Square to Difference of Proportions -- An Example: Chi-Square for Regions and Starts -- Case Study: Comparing Two Recommendation Systems with an A/B Test.

First Metric: Participating Sessions -- Second Metric: Daily Revenue Per Session -- Third Metric: Who Wins on Each Day? -- Fourth Metric: Average Revenue Per Session -- Fifth Metric: Incremental Revenue Per Customer -- Data Mining and Statistics -- No Measurement Error in Basic Data -- A Lot of Data -- Time Dependency Pops Up Everywhere -- Experimentation Is Hard -- Data Is Censored and Truncated -- Lessons Learned -- Chapter 5 Descriptions and Prediction: Profiling and Predictive Modeling -- Directed Data Mining Models -- Defining the Model Structure and Target -- Incremental Response Modeling -- Model Stability -- Time-Frames in the Model Set -- Prediction Models -- Profiling Models -- Directed Data Mining Methodology -- Step 1: Translate the Business Problem into a Data Mining Problem -- How Will Results Be Used? -- How Will Results Be Delivered? -- The Role of Domain Experts and Information Technology -- Step 2: Select Appropriate Data -- What Data Is Available? -- How Much Data Is Enough? -- How Much History Is Required? -- How Many Variables? -- What Must the Data Contain? -- Step 3: Get to Know the Data -- Examine Distributions -- Compare Values with Descriptions -- Validate Assumptions -- Ask Lots of Questions -- Step 4: Create a Model Set -- Assembling Customer Signatures -- Creating a Balanced Sample -- Including Multiple Timeframes -- Creating a Model Set for Prediction -- Creating a Model Set for Profiling -- Partitioning the Model Set -- Step 5: Fix Problems with the Data -- Categorical Variables with Too Many Values -- Numeric Variables with Skewed Distributions and Outliers -- Missing Values -- Values with Meanings That Change over Time -- Inconsistent Data Encoding -- Step 6: Transform Data to Bring Information to the Surface -- Step 7: Build Models -- Step 8: Assess Models -- Assessing Binary Response Models and Classifiers.

Assessing Binary Response Models Using Lift -- Assessing Binary Response Model Scores Using Lift Charts -- Assessing Binary Response Model Scores Using Profitability Models -- Assessing Binary Response Models Using ROC Charts -- Assessing Estimators -- Assessing Estimators Using Score Rankings -- Step 9: Deploy Models -- Practical Issues in Deploying Models -- Optimizing Models for Deployment -- Step 10: Assess Results -- Step 11: Begin Again -- Lessons Learned -- Chapter 6 Data Mining Using Classic Statistical Techniques -- Similarity Models -- Similarity and Distance -- Example: A Similarity Model for Product Penetration -- The Business Problem -- Data Used for Similarity Model -- Steps for Building a Similarity Model -- Step 1: What Distinguishes High Penetration Towns from Low Penetration Towns? -- Step 2: What Would the Ideal Town Look Like? -- Step 3: How Far Is Each Town from the Ideal? -- Evaluating the Similarity Model -- Table Lookup Models -- Choosing Dimensions -- Partitioning the Dimensions -- From Training Data to Scores -- Handling Sparse and Missing Data by Removing Dimensions -- RFM: A Widely Used Lookup Model -- RFM Cell Migration -- RFM and the Test-and-Measure Methodology -- Every Campaign Is an Experiment -- New Types of Campaigns Should Be Tested Before Being Rolled Out -- RFM and Incremental Response Modeling -- Naïve Bayesian Models -- Some Ideas from Probability -- Probability, Odds, and Likelihood -- Converting for Convenience -- The Naïve Bayesian Calculation -- Comparison with Table Lookup Models -- Linear Regression -- The Best-fit Line -- Goodness of Fit -- Residuals -- R2 -- Global Effects -- Multiple Regression -- The Equation -- The Range of the Target Variable -- Interpreting Coefficients of Linear Regression Equations -- Capturing Local Effects with Linear Regression.

Additional Considerations with Multiple Regression.