Feature Selection: supporting the mining process on cyber-physical systems result datasets

Cyber physical systems (CPs) often generated large sets of data during monitoring or testing processes. Analyzing these results manually is not practical as it requires great human effort. Machine learning can be a valuable approach to support the analysis and can help the responsible professional to make urgent decisions. Moreover, most of time these datasets contain missing, extreme, duplicate or defective values that can bias the general classification methods, which can be worked around with feature selection techniques. However, identifying all the possible combinations of features and select the best set of them is not an easy task. In this work, we present a feature selection study to automate the analysis of CPs test result datasets by the support of machine learning. The idea is to automatically identify a set of attributes that optimize the accuracy of the chosen machine learning model. Three scenarios that use large amounts of data from cyber physical systems were used and the results of feature selection were surprising in some cases.