Results of CPD methods experiments on TCPDBench

This data artifact contains the results of the experiments from a modified version TCPDBench CPD evaluation tool (here) in which we ran the hyper parameter tuning of CPD methods on 174 timeseries from Mozilla systems. The data is comprised of the following folders: offline_methods_experiments folder: it contains the experimental results of running the following offline CPD methods on the 174 timeseries: AMOC, BinSeg, BOCPD,CPNP, KCPA, MongoDB, PELT, RFPOP, SegNeigh, WBS, and the dummy method Zero. online_methods_experiments folder: it contains the experimental results of running the following online CPD methods on the 174 timeseries: ADWIN, BOCPD, Chisquare, CUSUM, CVMWIN, EWMA, KSWIN, MOSUM, and the dummy method ODummy. hybrid_methods_experiments folder: it contains the experimental results of running the following hybrid CPD methods on the 174 timeseries: CVM, KS, Levene, Welch, and MWU. mozilla_method_experiments folder: it contains the experimental results of running Mozilla's CPD method on the 174 timeseries. used_timeseries folder: It contains the 174 JSONfied timeseries to be plugged in directly to TCPDBench for the running the experiments. all_timeseries_attributes.json file: It contains the attributes of several timeseries, including the 174 timeseries brough up earlier. This file is not necessary for running TCPDBench experiments but it could help with analysis (for example grouping results per test, suite, platform, etc). annotations_no_variance.json and annotations_with_variance.json files: They are the annotations files containing the ground truth data based on which the experiments could be evaluated. In the annotations data collection process, we request annotators to label the annotations by their type (variance change, mean change, or both). Subsequent analysis led to the conclusion that variance annotaitons are to be eliminated to conduct the CPD methods assessment, so we use annotations_with_variance.json file for assessment. However, we include also the annotations file containing all annotation, including the vairance ones under annotations_no_variance.json file in case interesting findings could be generated out of it.