Documentation

Features

Here we present the main documentation of the pm4py library. It supports importing/exporting functionality of several different artifacts, basic process discovery and conformance checking functionalities, as well as computing different quality dimensions of process models.
Please use the navigation menu on the left to browse into the specific category of interest.

Importing

  • Importing event logs from IEEE XES files
  • Importing event logs from comma-separated-values (csv) files
  • Importing event logs from Parquet files
  • Import Petri nets with initial and final marking from PNML files

Exporting

  • Exporting event logs to IEEE XES files
  • Exporting event logs to csv files
  • Exporting event logs to Parquet files
  • Exporting Petri nets with initial and final marking to PNML files

Interval logs

  • Business Hours module
  • Lead Time and Cycle Time: incremental calculation
  • Calculation of the time passed from the previous activity
  • Filtering on activities duration (from START to COMPLETE duration)
  • (in the predictionIntegration branch) Usage of the business hours in the prediction

Process Discovery

  • Discovery of Directly-Follows Graphs (including frequency and performance decoration)
    • Conversion of Directly-Follows Graphs to workflow nets
  • Discovery of Petri nets with the Alpha Miner
  • Discovery of Petri nets with the Inductive Miner (DFG-Based)
  • Heuristics Miner: Discovery of Heuristics Net from log objects
  • Heuristics Miner: Conversion into Petri net

Replay/Conformance Checking techniques

  • Token-based replay
    • Diagnostics based on token-based replay (throughput analysis and root cause analysis)
  • Alignments-based replay
    • Visualization of the alignments result on top of a Petri net (beta; available in the ‘develop’ branch)
    • Visualization of the alignments result on top of a BPMN diagram (beta; available in the ‘bpmnIntegration2’ branch)
    • Visualization of the alignments table (beta; available in the ‘develop’ branch)

Process model quality evaluation

  • Replay fitness
  • Precision
  • Generalization
  • Simplicity

Filtering

  • Filtering logs:
    • Filtering by timeframe (containing, intersecting, keep events)
    • Filtering by case performance
    • Filtering by start activities
    • Filtering by end activities
    • Filtering by variants
    • Filtering by string attributes
    • Filtering by numeric attributes
    • Filtering by paths
  • Filtering Pandas dataframes:
    • Filtering by timeframe (containing, intersecting, keep events)
    • Filtering by case performance
    • Filtering by start activities
    • Filtering by end activities
    • Filtering by variants
    • Filtering by string attributes
    • Filtering by numeric attributes

LTL Checking

  • LTL Checking on logs
  • LTL Checking on Pandas dataframes

Case Management:

  • Statistics on variants
  • Statistics on cases
  • Getting events for a case given case ID

Graphs:

  • Graphicating logs:
    • Case duration
    • Events over time
    • Numeric attribute
  • Graphicating Pandas dataframes:
    • Case duration
    • Events over time
    • Numeric attribute

Big CSV management (Pandas):

  • Loading of a sample
  • Filtering of the dataframe
  • Lazy conversion of the timestamp columns
  • Obtaining the Directly-Follows Graph from the Pandas dataframe
  • Filtering the DFG given a noise threshold
  • Retrieving aggregated statistics without replay

Log management

  • Sampling
  • Sorting
  • Classifiers

Petri net analysis

  • Soudness analysis
  • Discovery of cycles
  • Discovery of strongly-connected components

Social Network Analysis:

  • Handover of Work metric
  • Working Together metric
  • Subcontracting metric
  • Similar Activities metric
  • Roles detection (grouping activities having similar sets of resources)
  • Decision Trees
  • Decision tree about the ending activity of a process
  • Decision tree about the duration of a case (Root Cause Analysis)

Stochastic Petri nets:

  • Building Stochastic Map for Petri net transitions
  • Tangible Reachability Graph from Stochastic Petri nets
  • Continuous-Time Markov Chains:
    • Transient Analysis
    • Steady-State Analysis
  • Linear Programming Performance Bounds

BETA FEATURES:

Decision Trees

  • Trace Clustering
  • Decision tree about the duration of a path leading to an activity (Root Cause Analysis)
  • Concept Drift Detection

BPMN 2.0 Management:

  • Importing/Exporting BPMN 2.0 diagrams (without layouting)
  • Representation of BPMN diagrams with curved arcs
  • Conversion of a Petri net into a BPMN model
  • Transforming a decorated Petri net into a decorated BPMN model
  • Conversion of a BPMN model to a Petri net:
    • Decoration purposes
  • Basic BPMN layouting on XML

Prediction of the Remaining Time:

  • Prediction through ElasticNets
  • Prediction through Keras RNN