VariantAnalysis

Variant Analysis based on Declarative Specifications

This is the base command to launch Janus variant analysis:

`java -cp Janus.jar $JANUS_MAINCLASS -iLF1 ${INPUT_LOG_1} -iLE1 ${ENCODING_INPUT_LOG_1} -iLF2 ${INPUT_LOG_2} -iLE2 ${ENCODING_INPUT_LOG_2} -oCSV ${OUTPUT_PATH}

where:

• JANUS_MAINCLASS is "minerful.JanusVariantAnalysisStarter"
• INPUT_LOG_1 and INPUT_LOG_2 the paths to the input log files
• INPUT_LOG_1_ENCODING and INPUT_LOG_2_ENCODING the encodings of the input event logs (default is "xes", alternatives are "strings" for TXT logs)
• OUTPUT_FILE the path where to save the results as CSV files

To show the list of all commands use: java -cp Janus.jar $JANUS_MAINCLASS -h

Parameters

-h,--help print help

Variant analysis specific parameters:

-iLF1,--in-log-1-file <path> path to read the log file from

-iLE1,--in-log-1-encoding <language> input encoding language {'xes','mxml','strings'}. Default is: 'xes'

-iLE2,--in-log-2-encoding <language> input encoding language {'xes','mxml','strings'}. Default is: 'xes'

-iLF2,--in-log-2-file <path> path to read the log file from

-oCSV,--out-csv-file <path> path to output CSV file

-permutations,--number-of-permutations <number> number of permutations to perform during the statistical test. default: 1000

-pValue,--p-value <number> p-value threshold for statistical relevance of the results. default: 0.01

-pValueAdjustment,--p-value-adjustment-method <language> pValue adjustment methods: do nothing, Holm-Bonferroni, Benjamini-Hochberg {'none','hb','bh'}. Default is: 'hb'

-measure,--measure <name> measure to use for the comparison of the variants. default: Confidence

-measureThreshold,--measure-threshold <number> threshold to consider the measure relevant. default: 0.0

-differenceThreshold,--difference-threshold <number> threshold for the difference of the variants constraints measurement to be considered relevant, default: 0.01

-simplify,--simplification-flag Flag if the output rules set shoul dbe simplified according to rules hierarchy. Default: false

-bestNresults,--number-of-best-results <number> Number of rules to return in among the best results. Default: 10

-oKeep,--output-keep keep irrelevant results in output

If declarative specifications are available:

-iMF1,--in-model-1-file <path> path to read the file of input model of log 1

-iME1,--in-model-1-encoding <language> input encoding language {'declare-map','minerful','json'}. Default is: 'json'

-iMF2,--in-model-2-file <path> path to read the file of input model of log 2

-iME2,--in-model-2-encoding <language> input encoding language {'declare-map','minerful','json'}. Default is: 'json'

If declarative specifications not available, Discovery Parameters:

-s,--support <threshold> threshold for support (reliability); it must be a real value ranging from 0.0 to 1.0. Default is: '0.95'

-c,--confidence <threshold> threshold for confidence level (relevance); it must be a real value ranging from 0.0 to 1.0. Default is: '0.25'

-oModel1CSV,--save-model-1-as-csv <path> save discovered model 1 in CSV format into the specified file

-oModel1JSON,--save-model-1-as-json <path> save discovered model 1 in JSON format into the specified file

-oModel2CSV,--save-model-2-as-csv <path> save discovered model 2 in CSV format into the specified file

-oModel2JSON,--save-model-2-as-json <path> save discovered model 2 in JSON format into the specified file

OUTPUT

The output is composed of a CSV file with the full detailed results of the program and a TXT file with the best N results in natural language.

The detailed CSV presents the following data:

• Constraint: name of the declarative rule
• p_value: p-value of the rule after the permutation test
• Measure_VAR1: measurement of the rule in LOG1
• Measure_VAR2: measurement of the rule in LOG2
• Absolute Difference: absolute difference between the rule measurements between the logs
• Natural_Language_Description: natural language description of the result

The TXT file presents the natural language description of the best N results of the CSV files ordered by their absolute difference in decreasing order.

REMARKS

• If no models are given in input for the event logs, the discovery component will be automatically launched to retrieve such models.

Variant Analysis based on DFG

This is the base command to launch Janus variant analysis:

`java -cp Janus.jar $JANUS_MAINCLASS -iLF1 ${INPUT_LOG_1} -iLE1 ${ENCODING_INPUT_LOG_1} -iLF2 ${INPUT_LOG_2} -iLE2 ${ENCODING_INPUT_LOG_2} -oCSV ${OUTPUT_PATH}

where:

• JANUS_MAINCLASS is "minerful.JanusDFGVariantAnalysisStarter"
• INPUT_LOG_1 and INPUT_LOG_2 the paths to the input log files
• INPUT_LOG_1_ENCODING and INPUT_LOG_2_ENCODING the encodings of the input event logs (default is "xes", alternatives are "strings" for TXT logs)
• OUTPUT_FILE the path where to save the results as CSV files

To show the list of all commands use: java -cp Janus.jar $JANUS_MAINCLASS -h

Parameters

-h,--help print help

Variant analysis specific parameters:

-iLF1,--in-log-1-file <path> path to read the log file from

-iLE1,--in-log-1-encoding <language> input encoding language {'xes','mxml','strings'}. Default is: 'xes'

-iLE2,--in-log-2-encoding <language> input encoding language {'xes','mxml','strings'}. Default is: 'xes'

-iLF2,--in-log-2-file <path> path to read the log file from

-oCSV,--out-csv-file <path> path to output CSV file

-permutations,--number-of-permutations <number> number of permutations to perform during the statistical test. default: 1000

-pValue,--p-value <number> p-value threshold for statistical relevance of the results. default: 0.01

-pValueAdjustment,--p-value-adjustment-method <language> pValue adjustment methods: do nothing, Holm-Bonferroni, Benjamini-Hochberg {'none','hb','bh'}. Default is: 'hb'

-differenceThreshold,--difference-threshold <number> threshold for the difference of the variants constraints measurement to be considered relevant, default: 0.00

-oKeep,--output-keep keep irrelevant results in output

OUTPUT

The output is composed of a CSV file with the full detailed results of the program.

The detailed CSV presents the following data:

• FROM-TO: name of the transition initiation and destination Events
• PERSPECTIVE: if the difference pertains the Average, Maximum, or Minimum time of this transition
• p_value: p-value of the rule after the permutation test
• Difference: absolute difference between the transition time between the logs
• LOG1-LOG2: original measurements in the respective variants