Loan Payback Predictor

Technologies:

Deep Learning

TensorFlow/Keras

Classfication

View Code

Goal

To predict whether or not a borrower will pay back their loans.

Process

Built a sequential DL model with keras to predict whether or not a borrower will pay back their loans with an 89% accuracy. The model consisted of 4 layers: 1 input layer, 2 hidden layers, 1 output layer. Since this is a classification problem, sigmoid activation function was used for the output layer. The other layers has a ReLu (Rectified Linear Unit) Activation functions. Dense, dropout layers. The loss function used was binary crossentropy and the optimiser used was Adam.