Replication Package: Anomaly detection via runtime monitoring data for structural equation modeling

<p>This replication package contains the following information:</p> <ul> <li><strong>Data extraction from literature &amp; interviews: </strong><em>Generation Structural &amp; Measurement Model via literature and interviews.xlsx</em> - here you can find the mapping of the extracted phrases to inductively summarise information regarding the structural and measurement models.</li> <li><strong>Dataset</strong> of runtime monitoring data extracted from TrainTicket via EvoMaster:&nbsp;<br> <ul> <li><em>TrainTicket faults classification.xlxs:</em> Describes the datasets and their faults, in which microservice the fault is injected for better explainability of the obtained results</li> <li><em>IndicatorDescriptionbasedonAnomalyDetectionToolsInterviews.xlsx:</em> description and mapping of selected indicators to the defined parameters from <a href="https://arxiv.org/abs/2408.07816" target="_blank" rel="noopener">previous work&nbsp;</a></li> <li>Unfortunately, the size of the datasets generated via EvoMaster and their injected faults are too big to upload here, thus, they will be available here: <a href="https://uibkacat-my.sharepoint.com/:f:/g/personal/monika_steidl_uibk_ac_at/EjLMt8SYWwtJtp2YuSaqavcBKJoCQ3b5H_l_OY0ifbVRCA?e=fatKyD" target="_blank" rel="noopener">Datasets with injected anomalies</a><br> <ul> <li>the error description can be found <a href="https://github.com/FudanSELab/train-ticket/wiki/Fault-Description" target="_blank" rel="noopener">here</a></li> <li>the datasets are named ts-error-<em>indicatorOfError</em>-reset.zip because the databases are getting reset so that no anomalies are introduced with wrong database entries</li> </ul> </li> </ul> </li> <li><strong>Code</strong> for handling and transforming data to extract indicators describing the whole system's and microservices' behavior from the collected runtime monitoring data collected from TrainTicket:<br> <ul> <li><a href="https://github.com/moniSt13/ConTest-Parsing" target="_blank" rel="noopener">link to the Github repository</a></li> </ul> </li> <li><strong>reports</strong> regarding the established PLS-SEM model using previously handled and transformed runtime monitoring data. Please be aware that opening the reports can leas to out of memory due to their size: <ul> <li><em>Assessment of Measurement Model: MeasurementModel_TrainTicket_erorcleaned.zip &amp; MeasurementModel_Bootstrap_ALL_TrainTicket_errorcleaned.zip</em>&nbsp;</li> <li><em>Assessment of Structural Model:&nbsp;StructuralModel_TrainTicket_errorcleaned.zip &amp; StructuralModel_Bootstrap_ALL_TrainTicket_errorcleaned.zip</em></li> </ul> </li> </ul> <p><br><br>---------------------------------------</p> <p><em>Future work </em>not elaborated in the associated paper due to space restrictions:</p> <ul> <li><strong>reports regarding F5 error</strong>: PLS-SEM model results without interpretation and further mediating effects between microservices included: F5_error.zip</li> </ul> <p>&nbsp;</p> <p>&nbsp;</p> <p>&nbsp;</p>