Data being used: CreditCardData

This data contains 10,127 observations and I will be predicting customer attrition based off 19 predictors

Split data into 70% and 30% training and validation data sets

Created a model using random forest to predict customer attrition

Variable Importance

Most important variables in attrition: total transaction count and total transaction amount.

Accuracy of model, tested on the validation data

## 
##  
##    Cell Contents
## |-------------------------|
## |                       N |
## | Chi-square contribution |
## |           N / Row Total |
## |           N / Col Total |
## |         N / Table Total |
## |-------------------------|
## 
##  
## Total Observations in Table:  3101 
## 
##  
##                    | val$predicted_Attrition_Flag 
## val$Attrition_Flag | Attrited Customer | Existing Customer |         Row Total | 
## -------------------|-------------------|-------------------|-------------------|
##  Attrited Customer |               400 |                71 |               471 | 
##                    |          1709.589 |           275.967 |                   | 
##                    |             0.849 |             0.151 |             0.152 | 
##                    |             0.928 |             0.027 |                   | 
##                    |             0.129 |             0.023 |                   | 
## -------------------|-------------------|-------------------|-------------------|
##  Existing Customer |                31 |              2599 |              2630 | 
##                    |           306.166 |            49.422 |                   | 
##                    |             0.012 |             0.988 |             0.848 | 
##                    |             0.072 |             0.973 |                   | 
##                    |             0.010 |             0.838 |                   | 
## -------------------|-------------------|-------------------|-------------------|
##       Column Total |               431 |              2670 |              3101 | 
##                    |             0.139 |             0.861 |                   | 
## -------------------|-------------------|-------------------|-------------------|
## 
##

Financial impact of the model

Predicted to leave = 445 * 100 = $44,500.
Customer retention has a 30% success rate = .30 * 408 = 122.4.
122.4 * $2,300 = $281,520
Net savings = $281,520 - $44,500 = $237,020