Credit scoring algorithms, which guess the probability of default, are the methods banks use to determine whether or not a loan should be granted. This model helps to improve the state of the art in credit scoring by predicting that someone will experience financial distress in the next two years.
The models will help in determining which features banks majorly use to decide when to give loan and when not to.
We took our dataset from Kaggle we have the data of 251503 people, which was divided into training, test and validation dataset.
We created a hybrid model using Random Forest, Logistic Regression, AdaBoosting, MultiLayer Perceptron and trained each of these models on our dataset and used maximum voting technique to obtain the final predictions.
Validation Accuracy for Hybrid Model - 0.99577
Testing Accuracy for Hybrid Model - 0.99572