CHAPTER 4

PREDICTIVE ANALYTICS

4.1 OVERVIEW

4.1.1 Predictive Modeling

Predictive analytics refers to a series of techniques concerned with making more informed decisions based on an analysis of historical data. These methods are used throughout the industries of science and business. For example, pharmaceutical companies use these techniques to assess the safety of potential drugs before testing them in human clinical trials, and marketing organizations use these models to predict which customers will buy a specific product.

This section will use the following example to illustrate the process of building and using predictive models. A telecommunications company wishes to identify and prioritize all customers they believe are likely to change to another service provider. In an attempt to avoid losing business, the company will offer new products or new service options to these customers. Over the years, the organization collected customer data monthly that included decisions to switch to another service provider. Table 4.1 is an example of the information collected on customers, where each observation relates to a specific customer in one month. The data includes a variable, churn, where a 1 indicates a switch to a new service provider and a 0 indicates a continuation of service. The table also includes the specific month the observation was recorded (month), the age of the customer (age), the annual income of the customer (income), the number of months they have been a customer ...

Get Making Sense of Data II: A Practical Guide to Data Visualization, Advanced Data Mining Methods, and Applications now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.