This repository contains materials from my final project for Johns Hopkins University's "Practical Machine Learning" course on Coursera.

For this project, I used accelerometer data to predict the type of physical exercise being performed. Training data came from human participants who wore accelerometers in various places while performing 1 of 5 exercises. I evaluated the performance of multiple classification methods in predicting exercise type on a test set; these included a decision tree, random forest, boosting, and linear discriminant analysis. I found that the random forest method was most accurate.

You can view the final report here!