This program deals with the titanic dataset available on kaggle. The goal of the program is to predict which passengers survived based on a number of criteria provided.
The program is divided into 2 files:
- preProcess.py - Pre-Processes the data converting categorical variables into desired forms and also dealing with missing data. Finally writes into .npy files to be used for training and testing.
- classify.py - trains a 3 layer neural net built using TensorFlow. and writes the predictions into csv file.
The classifier achieved a accuracy of 72%.
Skills Used: Python with numPy, pandas and TensorFlow.
- TensorFlow documentation
- https://beckernick.github.io/neural-network-scratch/