Bank customer classification

Main goal

Developing ML model predicting bank' customer inclination to open a deposit

Tasks

1. EDA;
2. Feature preprocessing;
3. Feature selection;
4. Data scaling;
5. Fitting Models (LogisticRegression, DecisionTree, RandomForrest); Hyperparameter optimization; Scores evaluation

Data set feature description

Customer details:

• age;
• job;
• marital (relationship status);
• education (level of education);
• default (has got an expired credit);
• housing (has got a housing loan);
• loan (has got a personal loan);
• balance.

Features related to the last contact:

• contact (contact type with a customer);
• month (month of the last contact);
• day (day of the last contact);
• duration (contact duration, seconds).

Other features:

• campaign (quantity of contacts with a client durint the current campaign);
• pdays (quantity of days missed since the last marketing campaign till the contact in the current campaign);
• previous (quantity of contacts till the current campaign)
• poutcome (the result of the previous campaign).

Target:

• deposit. Defines if a customer agrees to open a credit in a bank.

Summary

* Following classifiers were tested and compaired during the investigation: LogisticRegression, DecisionTree, RandomForest, GradientBoosting, Stacking
* DecisionTree, RandomForest, GradientBoosting models provided comparable scores after hyperparameters optimization  
* The best scores obtained with StackingClassifier after threshold optimizing: f1-score=0.83, accuracy=0.83

Tools used

pandas==2.0.1 numpy==1.23.5 matplotlib==3.6.3 optuna==3.3.0 seaborn==0.12.2 scikit-learn==1.3.0