Assignment 9

Objective

Create a random forest model to predict customer attrition using the CreditCardData data from the AER package

Data Prep

• Split the data (10127 observations), into training and validation datasets (70%/30% , 7089/3038)

Variable Importance

The total transaction count followed by total transaction amount is the greatest indicator for whether someone will be an attrited customer.

Model Accuracy

Using the validation data, it is observed that the model is able to:

• Predict correctly 83.9% of the existing customers
• Misclassify existing customers as an attrited customer 3.5% of the time

## 
##  
##    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 |               392 |                90 |               482 | 
##                    |          1558.549 |           253.999 |                   | 
##                    |             0.813 |             0.187 |             0.161 | 
##                    |             0.936 |             0.035 |                   | 
##                    |             0.131 |             0.030 |                   | 
## -------------------|-------------------|-------------------|-------------------|
##  Existing Customer |                27 |              2481 |              2508 | 
##                    |           299.530 |            48.815 |                   | 
##                    |             0.011 |             0.989 |             0.839 | 
##                    |             0.064 |             0.965 |                   | 
##                    |             0.009 |             0.830 |                   | 
## -------------------|-------------------|-------------------|-------------------|
##       Column Total |               419 |              2571 |              2990 | 
##                    |             0.140 |             0.860 |                   | 
## -------------------|-------------------|-------------------|-------------------|
## 
## 

Conclusions

• We offered to forgive 419 annual fees at $100 each (419 x $100 = $41900)
• 27 customers of the 419 were going to stay anyway. However, using our customer retention program we can keep 30% of the 392 customers that will actually leave
• This means that we will convince 118 customers to stay (30% x 392)
• The company makes $2300 per customer so through our program we keep $271400 (118 x $2300 = $271400)
• We net ($271400 - $41900) $229500.
• So for every 2990 customers we generate $229500
• We save approximately $77 per customer