How Much Event Data Is Enough? A Statistical Framework for Process Discovery

Bauer, M.; Senderovich, Arik; Gal, Avigdor; Grunske, Lars; Weidlich, Matthias

doi:10.1007/978-3-319-91563-0_15

Cited by 25 publications

(27 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Furthermore, to apply our proposed method on various event logs and use different process discovery algorithms with their different parameters, we ported the Sample Variant plug-in to RapidProM [3] which extends RapidMiner with process analysis capabilities. In our experiments, we also used the statistical sampling method that is presented in [8]; however, as we consider only work-flow information, its relaxation parameter is ignored.…”

Section: Methodsmentioning

confidence: 99%

“…However, we are not able to use this sampling technique for the process discovery purpose, because, the Parikh vector does not store the sequences of activities that are critical for discovering process models. In [8], the authors recommend a random trace-based sam- pling method to decrease the discovery time and memory footprint. This method assumes that process instances have different behavior if they have different sets of directly follows relations.…”

Section: Related Workmentioning

confidence: 99%

“…Here, we consider c = 10 and all instance selection strategies. Statistical corresponds to the sampling method that is proposed in [8] (with its default setting). It is clear that by reducing the size of an event log, the required process discovery time is reduced.…”

Section: Performance Analysismentioning

confidence: 99%

“…Even the basic Inductive miner that uses DF relations may result in different process models for event logs that have identical sets of DF relations. Figure 9 indicates that the average percentage of the remaining variants in sampled event logs using the statistical sampling method [8]. .…”

Section: Performance Analysismentioning

confidence: 99%

“…. Average of remained variants and directly follows relations in sampled event logs using the statistical sampling method [8].…”

Section: Performance Analysismentioning

confidence: 99%

See 4 more Smart Citations

Improving the performance of process discovery algorithms by instance selection

2020

View full text Add to dashboard Cite

Process discovery algorithms automatically discover process models based on event data that is captured during the execution of business processes. These algorithms tend to use all of the event data to discover a process model. When dealing with large event logs, it is no longer feasible using standard hardware in limited time. A straightforward approach to overcome this problem is to down-size the event data by means of sampling. However, little research has been conducted on selecting the right sample, given the available time and characteristics of event data. This paper evaluates various subset selection methods and evaluates their performance on real event data. The proposed methods have been implemented in both the ProM and the RapidProM platforms. Our experiments show that it is possible to considerably speed up discovery using instance selection strategies. Furthermore, results show that applying biased selection of the process instances compared to random sampling will result in simpler process models with higher quality.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Performance Analysismentioning

confidence: 99%

Section: Performance Analysismentioning

confidence: 99%

“…. Average of remained variants and directly follows relations in sampled event logs using the statistical sampling method [8].…”

Section: Performance Analysismentioning

confidence: 99%