Taxi Trip Time Deep Learning Model
A CSE 151B - Deep Learning Project
Taxi Trip Time Deep Learning Model
This deep learning model was built in collobaration with 3 other students under the group name GMNP. The task of the model is meant to predict the travel time of a taxi in Porto, Portugal given information about the taxi call including the time of the call, the type of call, the taxi stand if applicable and other such data. The data consisted of 1.7 million recorded taxi trips. This dataset was cleaned and useful features were extracted to create the model. The model ranked 21 out of 87 participating teams.
Built in Python using PyTorch
The model was built using the PyTorch Library. The source code for the models can be found here. The data cleaning, analysis and feature engineering were done using the Pandas library. The model was trained on UCSD's DataHub using CUDA.
