Bank customer Churn Analysis
Predicting customer churn in a bank thank to Machine Learning Interpretation Exited: Whether the customer has churned (1 = yes, 0 = no) The model shows an accuracy of 81% which is a good performance of the model The precision shows that out of all the customer that the model predicted would churned, only 66% actually did. The recall shows that out of all the customers that actually did churned, the model only predicted this outcome correctly for 19% of those customers. The f1-score shows that the model does a good job of predicting the unchurned customers. The support : we have 1581 of class 0 and 419 of class 1, which is a really unbalanced dataset.
Overall, this classifier or model is better to find class 0 over class 1. That is not in accordance of our goal to better predict a bank customer churn.
Apr 22, 2025
