This repository host different approaches developed to solve the challenge proposed in the Big Data Bowl 2020.
Winner Solution: reproduces the 1st place winner solution of the NFL Big Data Bowl 2020 kaggle competition.
Winner Solution - Pytorch: implementation of the NFL Big Data Bowl 2020 winner solution using graph neural networks (torch geometric)
Player Influence Area - CNN: exploits the idea of player influence area proposed in "Wide Open Spaces: A statistical technique for measuring space creation in professional soccer". The idea is to feed a CNN with an array of 22 images, where each image represents the influence area of each player.
Graph Convolutional Network: represents the traking data using a graph G(V, E), where V is the set of nodes, and E the set of edges. It only considers the rusher against the defensive team in the representation of the data.