In this presentation I’ll show you two correlation matrices corresponding to the data set from the final assignment of the course Practical Machine Learning of this specialization. I included a link to the final results. Here I just show the final predictors included in my final model.
First tidydata set to explore and choose relevant predictors. Originally I’ve 52 predictors but some of them are highly correlated.
Correlation plot after remove predictors which correlation was > 0.75. I used the findCorrelation function to get the variables more correlated. We went from 52 predictors to 31. And, as seen in the completed report in Rpubs, the accuracy of the final model was the same while reducing training time.