Analysis Notebook:

NYC Taxi Fare Prediction using Pytorch

Key Libraries: Pytorch, Matplotlib, Seaborn, Pandas, Numpy

Language: Python

Project Details:

In this project, we use PyTorch to predict the taxi fare prices for the NY taxi fare dataset provided by New York City Taxi Fare Prediction Kaggle competition . We built a TabularModel() class to handle both categorical and continous variables. A Fully Connected Network with 3 hidden layers of size [200,300,200] was used. ReLu activation function was used on each layer, the layer was normalized and then a dropout layer was used to randomly drop 40% of the data. The model was run for 300 epochs resulting in an RMSE of 3.24.