Quick Insights for Generalized Regression
Intended audience: END-USERS DATA SCIENCE DEVELOPERS
AO Easy Answers: 4.3
Overview
The Generalized Regression Quick Insight can only be generated as a scheduled task created by the Solution Developer using the Insight Composer in the AO Platform. The output from the generated Quick Insight will be added to the Insights page in Easy Answers.
Purpose
Generalized Regression is a method that models the relationship between variables, using different techniques for different types of data (e.g., binary, count, continuous). Generalized Linear Models (GLMs) are a flexible generalization of ordinary linear regression that allow for response variables that have error distribution models other than a normal distribution.
Business Example
A telecom company wants to predict whether a customer is likely to churn (i.e., leave the company) based on their service usage, contract type, customer service interactions, and billing information. The goal is to proactively retain customers at high risk of churning.
Results
A Generalized Linear Model gives the telecom company a data-driven method to:
Quantify which factors drive customer churn
Predict churn likelihood for individual customers
Inform retention strategies, pricing, and customer service policies
Data Sample
CustomerID | MonthlyCharges | Tenure | CustomerSupportCalls | ContractType | InternetService | PaperlessBilling | Churn |
|---|---|---|---|---|---|---|---|
C001 | 70.35 | 2 | 5 | Month-to-Month | Fiber | Yes | 1 |
C002 | 45.50 | 12 | 1 | One-Year | DSL | No | 0 |
C003 | 99.99 | 1 | 6 | Month-to-Month | Fiber | Yes | 1 |
C004 | 60.00 | 24 | 0 | Two-Year | DSL | No | 0 |
C005 | 55.75 | 6 | 2 | One-Year | None | Yes | 0 |
C006 | 80.20 | 3 | 4 | Month-to-Month | Fiber | Yes | 1 |
C007 | 65.00 | 36 | 0 | Two-Year | DSL | No | 0 |
C008 | 50.10 | 8 | 3 | One-Year | DSL | Yes | 0 |
C009 | 90.30 | 1 | 5 | Month-to-Month | Fiber | Yes | 1 |
C010 | 40.00 | 18 | 1 | Two-Year | None | No | 0 |