“…For this reason, Model 3 was modified by increasing the number of engine cycles. When additional calculations were made for training sets, created as previously but containing averaged data from more engine cycles (15,20,25,30,35,40), the accuracy after 10-fold cross valida-tion (run once in each case) was 93.9%, 94.3%, 96.2%, 96.1%, 96.2%, and 94.3%, respectively. At the 50th run of 10-fold cross validation (with almost identical software, Linux counterpart C5.0), the averaged values amounted to 93.9%, 95.7%, 97.1%, 96.2%, 95.1%, and 93.1%, respectively.…”