elviskoech / bank-churn-prediction-app

bank-churn-prediction-data-app

Required Files

setup.sh procfile requirement.txt

• App: Bank Churn Prediction Data App

Project Outline

• Problem Statement
• Dataset Information
• Feature Processing and Feature Engineering
• Machine Learning Model Development
• Evaluating the result/metrics
• Conclusion

Problem Statement

• Customer retention is one of the primary growth pillars for products with a subscription-based business model. Several bad experiences – or even one – and a customer may quit. And if droves of unsatisfied customers churn at a clip, both material losses and damage to reputation would be enormous.
• Customer churn (or customer attrition) is a tendency of customers to abandon a brand and stop being a paying client of a particular business.
• I used supervised machine learning classification approach to solve this problem and based on the number of target class I built a binary classifier type of ML model.

Dataset Information

• Data Source : Github
• Columns : 14
• Rows : 10000

Feature Processing and Feature Engineering

• I used SelectKBest and ExtraTreesClassifier from Sklearn library to find the best features

Machine Learning Model Development

• Using LogisticRegression ML Estimator our model had an accuracy score of 0.813(81.3%)
• I had to evaluate the model further using Classification report and Cross validation

Evaluating the result/metrics

• Cross Validation had an accuracy of 80.73%
• Comparing the logistic Regression model to:
  • Decision Tree Classifier
  • Random Forest Classifier
  • Support Vector Machine
  • K nearest Classifier
  • naive_bayes
• Using Classification Report to determine F1_Score of different models:
  • LR F1-score 0.5958600508740877
  • DT F1-score 0.6895821798155766
  • RF F1-score 0.7513784461152883
  • SVM F1-score 0.5210022107590273
  • NB F1-score 0.6412824619876383
  • KNN F1-score 0.6245016923566131
• Random Forest Classifier model performed well compared to other models
• To improve the accuracy of the Random Forest model I used RandomSearchCv to tune the hyperparameters:
  • Hence the randomised search cv on random forest classifier gave us better accuracy which is 86.25% and a std of 0.99% and wrong predictions made by the model are 374/2000

Conclusion

• To conclude, we can use these ML models to predict customer churn with a higher accuracy and metrics
• In general, it’s the overall customer experience that defines brand perception and influences how customers recognize value for money of products or services they use.
• The reality is that even loyal customers won’t tolerate a brand if they’ve had one or several issues with it.