Assigned Project: Bank Loans*
--> Formulate business model
--> Analyze data
--> Prepare for modelling (one hot encoding, normalization, feature selection, etc)
--> ML models applied:
Logistic Regression;
Decision Tree with Randomized Search CV;
Stratified k-fold cross validation;
Random Forest;
XGBoost Classifier;
--> ANN with ReLu activation and 2 dense layers
--> Cross validation for model tuning
--> Performance & evaluation metrics
Models for predicting whether or not an applicant should be accepted for a loan are widely used, and a very common implementation of machine learning practices. This dataset contains 100k instances of data and 21 features, and the goal is to predict which individuals will be accepted for a loan. Note that there are important fairness considerations for this problem in making sure our model is not discriminatory and as such measures for dealing with imbalanced dataset have also been taken into account.
Final file with 12mb due to 4 interactive Plotly graphs.
Dataset: https://novasbe365- my.sharepoint.com/:f:/g/personal/sabina_zejnilovic_novasbe_pt/ElbyDvjH1EpFl Mn5ah6RmcUBfW3nnmcUh4sWpBQ5EnXEsA?e=A4gRKN