Some ML fun with F1 datasets.
The "process_data.ipynb" notebook contains the cleaning of the original datasets.
The "NN_v1_k.ipynb" contains the Keras implementation of several feed-forward NNs that aim at predicting the final championship position of a driver based on descriptors of their race results such as average final race position and standard deviation.
The "NN_v1_torch.ipynb" contains the Pytorch implementation of several feed-forward NNs that aim at predicting the final championship position of a driver based on descriptors of their race results such as average final race position and standard deviation.
I explore the impact of different choices of inputs, hyperparameters and architectures.