This repository contains public code to beat the benchmark in the Otto Group Product Classification Challenge on Kaggle.
Leader Board Score ~ 0.76
Run Time ~ 3 minutes on my laptop (i5)
The network contains 93 neurons in the input layer (corresponding to each feature), 3 neurons in the hidden layer and 9 neurons in the output layer (corresponding to each class). Increasing the number of neurons in the hidden layer and tuning other parameters gives a better LB score.