Logistic Regression Model Analysis

Our initial approach to predict customer churn involved utilizing a Logistic Regression model, with 'Exited' as the target feature.

Data Cleaning

Data preprocessing included the removal of N/A values and the conversion of categorical variables such as 'Gender' and 'Geography' into binary indicators.

Model Training and Results

The Logistic Regression model was configured and trained with the following parameters:

An accuracy score of approximately 78.52% was attained.

The visual representations of the model's performance are as follows:

Confusion Matrix
Confusion Matrix
Classification Report
Classification Report
Feature Importances
Feature Importances

Findings

While the accuracy is moderate, the classification report and confusion matrix suggest the model's predictive performance for class '1' (customers that leave) could be improved. The feature importances graph provides insight into which features most significantly influence the model's predictions, offering a direction for further analysis.