
416 Empirical Research in Software Engineering
MMS data set ranged from 72% to 76%, the precision values from 50% to 54%, the F-measure
values from 59% to 61%, and the AUC % values from 79% to 81%.
Figure11.7shows the top six performing models on Android Telephony data set. The
techniques used for developing these models were LR, SVM, LB, NB, RBF, and MLP. The
change prediction model developed by the LR technique gave the best results with an
accuracy value of 69%, a precision value of 77%, a F-measure value of 72%, and an AUC %
of 73%.
11.6.5 Hypothesis Testing and Evaluation
This section states the results of Friedman statistical test using ...