Assignment 9

The Objective

Creating a Random Forest Model To Predict Customer Attrition as pertaining to Credit Cards

The Data

The data has 10,127 observations. It was split into training and validation datasets at a ratio of (70/30)

The Model

The algorithm used was a random forest, predicting Attrition, using all other 19 predictors.

Variable Importance

Here the two most important variables are Total_Trans_Ct & Total_Trans_Amt, a deduction from these variables is that Credit Card Attrition is largely in relation to transaction costs and frequency of transactions to the Card, where variables such as Education Level hold no significance.

Confusion Matrix / Results

## 
##  
##    Cell Contents
## |-------------------------|
## |                       N |
## | Chi-square contribution |
## |           N / Row Total |
## |           N / Col Total |
## |         N / Table Total |
## |-------------------------|
## 
##  
## Total Observations in Table:  2990 
## 
##  
##                    | val$predicted_Attrition_Flag 
## val$Attrition_Flag | Attrited Customer | Existing Customer |         Row Total | 
## -------------------|-------------------|-------------------|-------------------|
##  Attrited Customer |               395 |                87 |               482 | 
##                    |          1566.305 |           258.102 |                   | 
##                    |             0.820 |             0.180 |             0.161 | 
##                    |             0.934 |             0.034 |                   | 
##                    |             0.132 |             0.029 |                   | 
## -------------------|-------------------|-------------------|-------------------|
##  Existing Customer |                28 |              2480 |              2508 | 
##                    |           301.020 |            49.603 |                   | 
##                    |             0.011 |             0.989 |             0.839 | 
##                    |             0.066 |             0.966 |                   | 
##                    |             0.009 |             0.829 |                   | 
## -------------------|-------------------|-------------------|-------------------|
##       Column Total |               423 |              2567 |              2990 | 
##                    |             0.141 |             0.859 |                   | 
## -------------------|-------------------|-------------------|-------------------|
## 
## 

Analysis of the Retention Program

• 423 customers where forgiven their annual fee of $100

• The program cost $42,300

• 395 were going to leave and were predicted too leave

• Assuming a 30% effectiveness, 118.5 customers

• The value of a customer is $2,300

• The gross profit is 118.5*2300 = $272,500

• The net profit of the program is $230,250 for 2,990 customers

• The net profit per customer is $77

• While A 30% Effectiveness can be considered to be substantial/Unrealistically high, a 4.65% effectiveness, calculated by multiplying 0.4653952300 equals the break even point of $42,300. Meaning any effectiveness percentage exceeding 4.65% would be considered profitable