Machine Learning Techniques for Gait Analysis in Skiing

We investigate the use of supervised machine learning on data from ski-poles equipped with force sensors, with the goal of auto- matically identifying which sub-technique the skier is using. Our first contribution is a demonstration that sub-technique identification can be done with high accuracy using only sensors in the pole. Secondly, we also compare different machine learning algorithms (LSTM neural networks and random forests) and highlight their respective strengths and weaknesses, providing practitioners working with sports data some guidance for choice of machine learning algorithms