BAM Assignment 9
Chris Millerick & Zack Maloy
2024-04-18
Data Being Used
- Number of Observations
- 10,127 Observations
- What we are Predicting
- Attrition
- Number of predictors
- 19 Predictors
Algorithm Used
- Random Forest
Our data was split into Attritted Customers and Existing Customers
## Attrited Customer Existing Customer
## 1627 8500Variable Importance
## Attrited Customer Existing Customer
## Customer_Age 0.0060591236 5.410383e-03
## Gender 0.0073908994 1.025958e-03
## Dependent_count 0.0010542983 7.681695e-04
## Education_Level -0.0002182911 -7.572138e-05
## Marital_Status 0.0004079267 9.608432e-04
## Income_Category 0.0015632415 3.104565e-04
## Card_Category 0.0001404961 2.903018e-04
## Months_on_book 0.0040229985 2.071285e-03
## Total_Relationship_Count 0.0251896824 3.009392e-02
## Months_Inactive_12_mon 0.0258730922 1.036779e-03
## Contacts_Count_12_mon 0.0138721587 4.655136e-03
## Credit_Limit 0.0177452440 1.710369e-02
## Total_Revolving_Bal 0.0650076704 5.066380e-02
## Avg_Open_To_Buy 0.0168509095 1.627478e-02
## Total_Amt_Chng_Q4_Q1 0.0349782407 6.900299e-03
## Total_Trans_Amt 0.1250391724 9.522885e-02
## Total_Trans_Ct 0.2247137766 9.463965e-02
## Total_Ct_Chng_Q4_Q1 0.0813746295 1.305941e-02
## Avg_Utilization_Ratio -0.0008575924 4.257756e-02
## MeanDecreaseAccuracy MeanDecreaseGini
## Customer_Age 0.0055129194 64.981239
## Gender 0.0020491961 16.958639
## Dependent_count 0.0008144732 24.319704
## Education_Level -0.0001007033 22.023302
## Marital_Status 0.0008719689 17.051191
## Income_Category 0.0005127817 20.322761
## Card_Category 0.0002676244 4.649054
## Months_on_book 0.0023870218 46.242924
## Total_Relationship_Count 0.0293009666 123.081931
## Months_Inactive_12_mon 0.0050455272 46.853543
## Contacts_Count_12_mon 0.0061491044 56.725801
## Credit_Limit 0.0172107569 67.422677
## Total_Revolving_Bal 0.0529427309 205.705174
## Avg_Open_To_Buy 0.0163794723 57.721650
## Total_Amt_Chng_Q4_Q1 0.0114274203 120.557125
## Total_Trans_Amt 0.1000506057 369.263671
## Total_Trans_Ct 0.1155976422 308.125899
## Total_Ct_Chng_Q4_Q1 0.0240787883 220.683639
## Avg_Utilization_Ratio 0.0355525979 116.550664
Accuracy of the 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: 3036
##
##
## | val$predicted_Attrition_Flag
## val$Attrition_Flag | Attrited Customer | Existing Customer | Row Total |
## -------------------|-------------------|-------------------|-------------------|
## Attrited Customer | 389 | 98 | 487 |
## | 1556.394 | 247.122 | |
## | 0.799 | 0.201 | 0.160 |
## | 0.935 | 0.037 | |
## | 0.128 | 0.032 | |
## -------------------|-------------------|-------------------|-------------------|
## Existing Customer | 27 | 2522 | 2549 |
## | 297.357 | 47.214 | |
## | 0.011 | 0.989 | 0.840 |
## | 0.065 | 0.963 | |
## | 0.009 | 0.831 | |
## -------------------|-------------------|-------------------|-------------------|
## Column Total | 416 | 2620 | 3036 |
## | 0.137 | 0.863 | |
## -------------------|-------------------|-------------------|-------------------|
##
## Financial Impact
446 people that we predicted were going to leave
Assuming that we gave every customer $100 that we thought were going to leave
The cost of our program 446 x 100 = $44,600
We will save 30% of 422 = 126.6, which has a value of $2,300 x 126.6 = $291,180
The net of this program is $246,580 from 3,017 customers
Per customer is $81.73