Practical_Machine_Learning

Weightlifting Prediction Analysis Using PCA and Random Forests

This is the Practical Machine Learning Project Repository. If you want a web-based version of the project go to this RPubs page. The basic outline of my solution is:

1. Clean the training and testing datasets
   • Remove non-relevant variables
   • Remove variables with more than 95% NAs
2. Do a Principal Component Analysis (PCA) on the clean training dataset in order to reduce the number of variables
3. Select the PCs whose cumulative proportion of explained variance equals 95%
4. Predcit the new PCs on the training dataset
5. Use the predictions to build a Random Forest model (parallelization is required)
6. Do a confusion matrix to assess the performance on the training dataset
7. Predict the class outcome on the testing set, first calculating the PCs and then using the Random Forest model