XGBoost and XGBoost Tuned Models for Predicting Customer Churn

The questions we are asking:

How can we keep customers from leaving and going to a new service provider? What are customers most concerned about in their decision to leave the company?

This exhibition utilizes an XGBoost and a Tuned XGBoost, the Tuned XGBoost is focused on the most significant variables, and the results for both are below.

Results With the XGB and XGB Tuned Models:

With our XGB Model tuned up there was a slight decrease in accuracy
from 80.33% to 79.62%,
A -0.71% difference.
There was a trade-off for a significant increase in sensitivity,
from 53.83% to 65.41%,
A 11.58% increase in ability to detect true positives.
Sensitivity = (True Positives)/(True Positives + False Negatives)

This translates into the abilty to predict if customers are going to leave, and the most significant reasons those customers are likely to leave. This will help reduce customer churn, and maintain/increase revenues.

## *************** XGB Train & Test: Accuracy & Sensitivity ***************

## Train accuracy:    82.83 % 
## Train sensitivity: 58.94 % 
## 
## Test  accuracy:    80.33 % 
## Test  sensitivity: 53.83 %

## ************ XGB Tuned Train & Test: Accuracy & Sensitivity ************

## Tuned Train accuracy:    81.35 % 
## Tuned Train sensitivity: 69.86 % 
## 
## Tuned Test  accuracy:    79.62 % 
## Tuned Test  sensitivity: 65.41 %

This Section will show any missing values in the data set in Orange.

## 
##  Variables sorted by number of missings: 
##          Variable Count
##            gender     0
##     SeniorCitizen     0
##           Partner     0
##        Dependents     0
##            tenure     0
##      PhoneService     0
##     MultipleLines     0
##   InternetService     0
##    OnlineSecurity     0
##      OnlineBackup     0
##  DeviceProtection     0
##       TechSupport     0
##       StreamingTV     0
##   StreamingMovies     0
##          Contract     0
##  PaperlessBilling     0
##     PaymentMethod     0
##    MonthlyCharges     0
##      TotalCharges     0
##             Churn     0

This section will show the most significant variables in the data set, the variables are measured by the quantity of churned customers.

The report was completed in 15.8 seconds.