Human Activity Recognition on Smartphones, understanding the dataset




I am trying to understand the data-set of the problem Human Activity Recognition on Smartphones which is available here->

It contains a training set with name X_train, as described by the readme file. In addition, it also contains a folder named inertial signals which again contains 9 txt files giving the data for different readings of the gyroscope and the accelerometer in the three dimensions. Which data set do we have to use to build our model for the problem, my guess is that X_train already contains the data of those 9 files, but I am not sure and not able to figure it out.