Process Discovery Algorithm Challenge 2020

Organized by Alexey - Current server time: Oct. 23, 2020, 9:43 p.m. UTC

Current

Development
Oct. 12, 2020, 6 a.m. UTC

Next

Evaluation
Nov. 1, 2020, 8:59 p.m. UTC

End

Competition Ends
Nov. 8, 2020, 8:59 p.m. UTC

Process Discovery Algorithm Challenge 2020

Participants should implement Heuristic Miner Algorithm.

All questions with regard to the challenge are discussed on the official Forum.

Evaluation

Evaluation is based on the following criteria:

  • equality to reference implementation made by pm4py - [0.0, 1.0]
  • performance - average duration of algorithm work on public\private logs
  • code quality by pylint normalized to [0.0, 1.0]

Leaderboard sudmissions are ranked by all three criteria with abovementioned ordering.

Submission to the competition involves uploading a .zip archive containing (see starting kit for sample submission):

  • metadata - empty file due to codalab requirements
  • miner.py - script with heuristic miner implementation algorithm

Starting kit contains sample submission.

Submission is run in docker container naydyonov/pm4py:version-1.5.2.2

All questions with regard to the challenge are discussed on the official Forum.

Rules

MIT licence is obligatory for code submited by participants.

Winner will receive a small token of appreciation.

All questions with regard to the challenge are discussed on the official Forum.

Starting Kit

Starting kit equips you with some template data and code compatible with the competition rules. We highly recommend reviewing the starting kit to elucidate the obscurities of code implementation or scoring procedure.

The starting kit comprises the folders:

  • sample_data. Here a log file resides for algorithm development purposes.
  • sample_reference. Example of heuristic net build by pm4py heuristic miner implementation over sample data.
  • sample_code_submission. Template for algorithm implementation. The task is to implement heuristic miner algorithm inside 'apply_hue_miner' function in 'sample_submission/miner.py' file.
  • ingestion. Here is the script which runs a code submitted. The script takes log file from sample data and puts the results into sample_result_submission.
  • scoring. A script which evaluates code submitted. We use exactly the same script for evaluation on private data. As you’re allowed to submit only a few times a day, it’s useful to have evaluation routine of your own.

Submission is run in docker container naydyonov/pm4py:version-1.5.2.2, so take in account please environment requirements.

The task is to implement heuristic miner algorithm inside 'apply_hue_miner' function in 'sample_submission/miner.py' file.

All questions with regard to the challenge are discussed on the official Forum.

Development

Start: Oct. 12, 2020, 6 a.m.

Description: Development phase: implement algorithm on public logs.

Evaluation

Start: Nov. 1, 2020, 8:59 p.m.

Description: Test phase: evaluation on private logs.

Competition Ends

Nov. 8, 2020, 8:59 p.m.

You must be logged in to participate in competitions.

Sign In
# Username Score
1 VitaLeeY 0.80
2 Sur_Daniil 0.45
3 aikimind 0.41