This is the repository for the NYU Intro to Data Science - 1001 final project.
Our objective for this project was to create a self driving algorithm on par with human players (experts) in the Car Racing environemnt from OpenAI Gym. We collected data by asking participants to play the game and try to get the best score they can in a number of trails. The screenshots of the track and the corresponding key presses were recorded as the car was being driven aroung the track. We employed machine learning (logistic regression, random forests) and deep learning (CNNs) to create models that could predict the actions required to drive cars around any given track. We achieved 75% and 99% accuracy on our best machine learning and best deep learning models respectively. Our paper can be found here.