Objective

 

Use machine learning technics including hierarchical clustering, K-means clustering, Naive-Bayes clustering and Random Forest clustering to predict segment membership and predict who is likely to subscribe to the cable service. 

 

Dataset Summary

​

Simulate customer segmentation data for a cable company so that 300 customers fall into 4 segments: "Suburb Mix", "Urban hip", "Travelers", "Moving-Up". Those segments are defined based on the following features:

• Age

• Gender

• Income

• Kids Count

• Home Owner or Renter

• Subscribed to the cable or not.

​

Content

​

1. Descriptive Data Analysis

   • Using command such as summary(), aggregate(), xtabs() to arrange and reshape dataset into the form desired for straightforward description.

2. Discreet Data Visualization

   • In histogram and bar chart

3. Continuous Data Visualization

   • In bar chart, boxplot and box-and-whiskers

4. Statistical Test

   • Chi-square test​

   • T-test: Testing Group Means

   • ANOVA: Testing Multiple Group Means

   • Bayes Statistics

5. Predict Membership and Segment Using:

   • Hierarchy Clustering​

   • K-Means Clustering

   • Naive-Bayes Clustering

   • Random Forest Clustering

​

Tools

​

R + RStudio

​

Library

 

library(cluster)

library(ggplot2)

library(factoextra)

library(mclust)

library(scatterplot3d)

library(MASS)

library(poLCA)

library(gplots)

library(e1071)

library(RColorBrewer)

library(psych)

​