Time series prediction using principal feature analysis

We give the formulation for time series prediction using principal feature analysis (PFA). PFA is a method introduced by Ira Cohen, Qi Tian et al. in 2002 for feature subset selection problem. PFA involves k-means formulation on significant principal components, and we adopt this PFA methodology for time series prediction. We demonstrate the usefulness of our formulation for the problem of prediction of machine tool wear. For this problem, we first construct our model using the first half of the time series data set, and we use the entire data set, including the second half, for the verification of our model. The verification is done by mean square error (MSE) criterion, and we demonstrate the selection of features for variable MSE, being in the range of 1.5% to 0.9% for the tool wear data set.