When putting the paraments for the train_test_split(), I am confused whether or not to put the design matrix of my x-data points or the matrix that only holds the original x-data points.
They both work, and do not output any errors, however which one is the correct one to use when doing a training/testing split?
(xcoord is the original data and X is the design matrix)
i.e:
X_train, X_test, y_train, y_test = train_test_split(xcoord, ydata, test_size=0.80)
vs
X_train, X_test, y_train, y_test = train_test_split(X, ydata, test_size=0.80)
Also, what is the difference, if there is any?