PM4Py Features - Algorithms and Capabilities

Process Model Types

PM4Py features: Process Model Types
Feature	Input	Added In	Scientific Publication	Examples
Petri Nets (PN)Workflow Nets (WF-nets)	–	2.0.0	Publication The Application of Petri Nets to Workflow Managementby van der Aalst	View
Maximal DecompositionSplits a net into smaller nets		2.0.0		View
Reachability GraphAll reachable markings of a net		2.0.0		View
Soundness Checking (WOFLAN)Checks whether a workflow net is sound		2.0.0	Publication Diagnosing workflow processes using Woflanby Verbeek et al.	View
Conversion from Process TreeTurns a process tree into a Petri net		2.0.0		View
Conversion from BPMNTurns a BPMN model into a Petri net		2.1.0		View
Conversion from POWLTurns a POWL model into a Petri net		2.7.5.2
Reset and Inhibitor ArcsArcs that empty a place or block a transition	–	2.2.10.2		View
Data Petri NetsTransitions guarded by data attributes	–	2.2.10.2		View
VisualizationWith frequency or performance decoration		2.0.0		View
Process TreeHierarchical, block-structured process model	–	2.0.0		View
Process Tree GeneratorRandom process trees for artificial data	–	2.0.0	Publication PTandLogGenerator: A generator for artificial event databy Jouck et al.	View
Conversion from Petri NetTurns a block-structured net into a tree		2.0.1	Publication Translating workflow nets to process trees: an algorithmic approachby van Zelst et al.	View
Conversion from POWLTurns a POWL model into a process tree		2.7.14.1
VisualizationOperators and activities as a tree		2.0.0		View
Business Process Model and Notation (BPMN)Import and export of BPMN 2.0 models	–	2.1.0	Publication Business Process Model and Notation (BPMN)by Object Management Group (OMG)	View
Conversion from Process TreeTurns a process tree into a BPMN model		2.1.0		View
Conversion from Petri NetTurns a workflow net into a BPMN model		2.1.0
VisualizationBPMN 2.0 diagrams		2.1.0		View
Partially Ordered Workflow Language (POWL)Process trees extended with partial orders	–	2.7.5.2
Conversion from Process TreeTurns a process tree into a POWL model		2.7.10
Conversion from Petri NetTurns a workflow net into a POWL model		2.7.14.1
VisualizationPOWL models with partial orders		2.7.5.2
Directly-Follows Graph (DFG)Which activity directly follows which		2.0.0	Publication A practitioner’s guide to process mining: Limitations of the directly-follows graphby van der Aalst	View
Filtering Activities and PathsKeeps the most frequent activities and paths		2.0.0		View
VisualizationWith frequency or performance decoration		2.0.0		View
Timeline DFGA DFG laid out along time		2.7.22	Publication Timeline-based Process Discoveryby Kaur, Mendling, Rubensson, and Kampik	View
Causal NetsActivities with input and output bindings		2.0.0	Publication Causal Nets: A Modeling Language Tailored towards Process Discoveryby van der Aalst et al.
Transition SystemStates and transitions abstracted from a log		2.0.0	Publication Process Mining - Data Science in Actionby van der Aalst	View
VisualizationStates and transitions as a graph		2.0.0		View
Prefix TreeRepresents all activity prefixes in a log	–	2.0.0
VisualizationThe activity prefixes of a log as a tree		2.2.5
Object-Centric DFG (OC-DFG)A DFG per object type, combined		2.2.16	Publication OC-PM: analyzing object-centric event logs and process modelsby Berti et al.	View
VisualizationObject-centric DFGs as a graph		2.2.16		View
Object-Centric Petri Nets (OC-PN)A Petri net per object type, combined		2.2.16	Publication Discovering object-centric Petri netsby van der Aalst et al.	View
VisualizationObject-centric Petri nets as a graph		2.2.16		View
Object-Centric Causal Nets (OC-CN)Causal nets across object types		2.7.20	Publication Object-Centric Causal Netsby Liss, Mensing, and van der Aalst	View

Event Data

PM4Py features: Event Data
Feature	Input	Added In	Scientific Publication	Examples
Event LogImport and export, with XES support	–	2.0.0		View
FilteringBy time, attributes, variants, paths, and more		2.0.0		View
Differential PrivacyAnonymizes a log while keeping its context		2.0.0	Publication PRIPEL: privacy-preserving event log publishing including contextual informationby Fahrenkrog-Petersen et al.	View
Label SplittingSplits activity labels by their context		2.7.9	Publication Context-Based Activity Label-Splittingby van Zelst et al.	View
Merging Related Logs (Case Relations)Two related logs merged into one		2.2.16		View
Object-Centric Event Log (OCEL)Supports the OCEL 1.0 and OCEL 2.0 standards	–	2.2.15	Publication The OCEL standard (1.0 and 2.0)	View
Filtering OCELBy object types, attributes, counts, and more		2.2.16		View
Statistics OCELObject counts, edge metrics, and more		2.2.18		View
Feature ExtractionFeature tables for objects and events		2.2.18	Publication Graph-based feature extraction on object-centric event logsby Berti et al.
OCEL OLAP OperationsDrill-down, roll-up, unfold, and fold		2.7.22.3	Publication Advancing Object-Centric Process Mining with Multi-Dimensional Data Operationsby Khayatbashi, Miri, and Jalali	View
FlatteningA traditional event log for one object type		2.2.15		View
ValidationChecks files against the OCEL schema		2.2.17		View
Object GraphsRelations between objects, as graphs		2.2.17.1		View
Timestamp-Based InterleavingsTwo related logs combined into one OCEL		2.2.15		View

Process Discovery

PM4Py features: Process Discovery
Feature	Input	Output	Added In	Scientific Publication	Examples
Alpha MiningPetri nets from directly-follows relations	–	–	2.0.0
Alpha MinerThe original algorithm			2.0.0	Publication Workflow Mining: Discovering Process Models from Event Logsby van der Aalst et al.	View
Alpha Miner +Adds support for short loops			2.0.0	Publication Workflow Mining: Current status and future directionsby Alves de Medeiros et al.	View
Inductive MiningSound, block-structured models from logs	–	–	2.0.0		View
Inductive Miner (IM)The base variant, with perfect fitness			2.0.0	Publication Discovering block-structured process models from event logs - A constructive approachby Leemans et al.	View
Inductive Miner Infrequent (IMf)Filters infrequent behavior (noise)			2.0.0	Publication Discovering block-structured process models from event logs containing infrequent behaviourby Leemans et al.	View
Inductive Miner DFG (IMd)Works on a DFG, for very large logs			2.0.0	Publication Scalable Process Discovery with Guaranteesby Leemans et al.	View
Heuristic MinerFrequency-based, robust to noise			2.0.0	Publication Flexible heuristics miner (FHM)by Weijters et al.	View
VisualizationWith dependency and frequency values			2.0.0		View
ILP MinerPetri nets via integer linear programming			2.6.0	Publication ILP-based process discovery using hybrid regionsby van Zelst et al.	View
Genetic MinerPetri nets by evolutionary search			2.7.20		View
Split MinerSimple, accurate BPMN models			2.7.22.4	Publication Split miner: automated discovery of accurate and simple business process models from event logsby Augusto et al.	View
DFG MinerTurns a DFG into a workflow net			2.0.0	Publication Directly Follows-Based Process Mining: Exploration & a Case Studyby Leemans et al.	View
Correlation MinerDiscovery without case identifiers			2.0.0	Publication Correlation miner: Mining business process models and event correlations without case identifiersby Pourmirza et al.	View
Prefix Tree ApproachPrefix tree of the traces in a log			2.0.0	Publication Skeletal algorithms in process miningby Przybylek
Declarative Process ModelsDeclare constraints between activities			2.7.6	Publication User-guided discovery of declarative process modelsby Maggi et al.	View
Log SkeletonOrdering and occurrence constraints			2.2.11	Publication The log skeleton visualizer in ProM 6.9by Verbeek	View
Partially Ordered Workflow Language (POWL) MinerDiscovers POWL models from logs			2.7.5.2	Publication POWL: partially ordered workflow languageby Kourani et al.	View
Local Process ModelsFrequent patterns in parts of the process			2.7.23.3	Publication Local Process Modelsby Tax et al.	View

Conformance Checking

PM4Py features: Conformance Checking
Feature	Added In	Scientific Publication	Examples
FootprintsCompares the relations of log and model	2.0.0	Publication Process Mining - Data Science in Actionby van der Aalst	View
VisualizationThe footprint matrix as a table	2.0.0		View
Token-Based ReplayReplays traces with tokens on a Petri net	2.0.0	Publication A novel token-based replay technique to speed up conformance checking and process enhancementby Berti et al.	View
Throughput and Root Cause AnalysisDurations and root causes of deviations	2.0.0		View
AlignmentsOptimal mapping of each trace to the model	2.0.0	Publication Conformance checking using cost-based fitness analysisby Adriansyah et al.	View
Approximate Alignments (Tandem Repeats)Compresses repeated patterns	2.7.23.2	Publication Efficient Conformance Checking using Approximate Alignment Computation with Tandem Repeatsby Reißner et al.	View
Approximate Alignments (Subset Selection and Edit Distance)Aligns a sample, estimates the rest	2.7.23.2	Publication Conformance Checking Approximation Using Subset Selection and Edit Distanceby Fani Sani et al.	View
Approximate Alignments (Sliding Window Top-k)Windowed alignment for very long traces	2.7.23.2	Publication A Scalable and Near-Optimal Conformance Checking Approach for Long Tracesby Bogdanov et al.	View
Approximate Alignments (Sequential Fixed Horizon)Aligns a few events at a time	2.7.23.2	Publication Aligning Modeled and Observed Behavior: A Compromise Between Computation Complexity and Qualityby van Dongen et al.	View
Alignments (decomposed)Aligns the parts of a decomposed net	2.0.0	Publication Recomposing conformance: Closing the circle on decomposed alignment-based conformance checking in process miningby Lee et al.	View
Anti-alignmentsModel runs far from the log, for precision	2.0.0	Publication Anti-alignments - measuring the precision of process models and event logsby Chatain et al.	View
Multi-alignmentsAligns several traces at once	2.0.0	Publication Optimized SAT encoding of conformance checking artefactsby Boltenhagen et al.	View
Alignments on a DFGAligns traces against a DFG	2.0.0		View
Alignments between Logs (Edit Distance)Matches the traces of one log to another	2.0.0		View
Process Tree Alignment ApproximatedFast approximation on process trees	2.0.0	Publication Alignment approximation for process treesby Schuster et al.
Process Tree Alignment MILPOptimal, via mixed-integer programming	2.0.0	Publication Process Tree Alignmentsby Schwanen et al.
Process Tree Alignment Dynamic ProgrammingOptimal, via dynamic programming	2.0.0	Publication A Dynamic Programming Approach for Alignments on Process Treesby Schwanen et al.
Extended Marking EquationHeuristic for fast optimal alignments	2.0.0	Publication Efficiently computing alignmentsby van Dongen
Alignment TableThe alignments of all variants as a table	2.7.4
Log SkeletonChecks a log against its constraints	2.0.0	Publication The log skeleton visualizer in ProM 6.9by Verbeek	View
Declare ConformanceChecks a log against Declare constraints	2.7.6
LTL CheckingRules such as the four-eyes principle	2.0.0	Publication Process mining and verification of properties: An approach based on temporal logicby van der Aalst et al.	View
Temporal ProfileFlags unusual times between activities	2.0.0	Publication Temporal Conformance Checking at Runtime based on Time-infused Process Modelsby Stertz et al.	View

Model Evaluation

PM4Py features: Model Evaluation
Feature	Input	Output	Added In	Scientific Publication	Examples
Replay FitnessHow much of the log the model can replay	–	–	2.0.0
Fitness (Token-Based Replay)Fast, from missing and remaining tokens			2.0.0		View
Fitness (Alignment-Based)Exact, from the alignment costs			2.0.0		View
PrecisionHow little extra behavior the model allows	–	–	2.0.0
Precision (ETConformance Token)Based on token-based replay			2.0.0	Publication A fresh look at precision in process conformanceby Munoz-Gama et al.	View
Precision (Align-ETConformance)Based on alignments			2.0.0	Publication Measuring precision of modeled behaviorby Adriansyah et al.	View
Precision (Automaton After Alignments)Builds an automaton from the alignments			2.0.0		View
Generalization (Token-Based)How well the model handles unseen behavior			2.0.0	Publication Quality dimensions in process discovery: The importance of fitness, precision, generalization and simplicityby Buijs et al.	View
SimplicityStructural complexity of the model	–	–	2.0.0
Simplicity (Arc Degree)Average number of arcs per node			2.0.0	Publications ProDiGen: Mining complete, precise and minimal structure process models with a genetic algorithmby Vázquez-Barreiros et al. Prediction of business process model quality based on structural metrics	View
Simplicity (Extended Cardoso)Complexity from splits and joins			2.0.0	Publication Complexity Metrics for Workflow Netsby Lassen and van der Aalst	View
Simplicity (Extended Cyclomatic)Complexity from the reachability graph			2.0.0	Publication Complexity Metrics for Workflow Netsby Lassen and van der Aalst	View
Earth Mover Distance (EMD)Compares log and model as distributions			2.0.0	Publication Stochastic process mining: Earth movers’ stochastic conformanceby Leemans et al.	View

Process Enhancement and Analysis

PM4Py features: Process Enhancement and Analysis
Feature	Added In	Scientific Publication	Examples
StatisticsDurations, arrivals, frequencies, and more	2.0.0		View
Case Arrival and DispersionHow often new cases start	2.0.0		View
Concurrent ActivitiesActivities running at the same time	2.0.2		View
Eventually-Follows GraphWhich activity eventually follows which	2.0.2		View
ReworkRepeated activities, per activity and case	2.2.8		View
Partially Ordered VariantsVariants that keep parallel events	2.7.23	Publication Defining and visualizing process execution variants from partially ordered event databy Schuster et al.	View
Batch DetectionActivities handled in batches	2.2.9	Publication Batch processing: Definition and event log identificationby Martin et al.	View
Process CubesTwo-dimensional slices of a log	2.0.0		View
Decision MiningAttribute rules behind the choices	2.0.0	Publication Data-aware process mining: discovering decisions in processes using alignmentsby de Leoni et al.	View
VisualizationTrees from decision mining or ML	2.0.0		View
Concept Drift DetectionChanges of the process over time	2.7.17	Publication Dealing with concept drifts in process miningby Bose et al.	View
Network AnalysisAggregated event links, for object-centric data	2.2.16		View
Link AnalysisEvent links in detail, for object-centric data	2.2.16		View

Organizational Mining

PM4Py features: Organizational Mining
Feature	Added In	Scientific Publication	Examples
Social Network AnalysisNetworks between the resources of a log	2.0.0	Publication Discovering social networks from event logsby van der Aalst et al.	View
Handover of WorkWho passes work to whom	2.0.0		View
SubcontractingWho works between two steps of another	2.0.0		View
Working TogetherWho works in the same cases	2.0.0		View
Similar ActivitiesWho performs similar activities	2.0.0		View
VisualizationResource networks as graphs	2.0.0		View
Organizational RolesGroups resources by their activities	2.0.0	Publication Business models enhancement through discovery of rolesby Burattin et al.	View
Resource ProfilesWorkload and behavior of each resource	2.2.4	Publication Mining resource profiles from event logsby Pika et al.	View
Group-Based Organizational MetricsMetrics on groups of resources	2.2.4		View

Process Simulation

PM4Py features: Process Simulation
Feature	Input	Added In	Scientific Publication	Examples
Monte Carlo SimulationSimulates cases with timing from the log		2.0.0		View
Play OutEvent logs from process models	–	2.0.0	Publication Process Mining - Data Science in Actionby van der Aalst
DFGEvent log generation from a DFG		2.0.0		View
Petri NetEvent log generation from a Petri net		2.0.0		View
Process TreeEvent log generation from a process tree		2.0.0		View

Machine Learning

PM4Py features: Machine Learning
Feature	Added In	Scientific Publication	Examples
Feature ExtractionFeature tables from the cases of a log	2.2.7.2		View
Temporal Feature ExtractionTime series of process indicators	2.3.4	Publication Supporting automatic system dynamics model generation for simulation in the context of process miningby Pourbafrani et al.	View
Trace EncodingsTraces as numeric vectors	2.7.22.4	Publication Trace encoding in process mining: A survey and benchmarkingby Tavares et al.	View
One-hot EncodingBinary occurrence vectors over trace tokens	2.7.22.4		View
Count2Vec EncodingTrace-token frequency count vectors	2.7.22.4		View
N-gram EncodingCounts of contiguous activity sequences	2.7.22.4		View
TF-IDF EncodingTF-IDF weighted token or n-gram vectors	2.7.22.4		View
Doc2Vec EncodingEach trace embedded as a document	2.7.22.4		View
Word2Vec EncodingAveraged activity embeddings	2.7.22.4		View
BERT EncodingSentence-transformer embeddings of traces	2.7.22.4		View
Token Replay EncodingToken replay diagnostics as features	2.7.22.4		View
Alignment EncodingAlignment diagnostics as features	2.7.22.4		View
Trace Clustering (hierarchical)Groups cases by their attributes	2.3.0	Publication A generic framework for attribute-driven hierarchical trace clusteringby van Zelst et al.
Predict Next ActivityLSTM model for predicting the next activity	2.0.0
Predict Remaining TimeLSTM model for the remaining cycle time	2.0.0

Streaming Process Mining

PM4Py features: Streaming Process Mining
Feature	Input	Output	Added In	Scientific Publication	Examples
Streaming ImportersRead logs as a stream of events	–	–	2.0.1
XES (Trace-by-Trace)Streams an XES file, trace by trace	–		2.0.1		View
XES (Event-by-Event)Streams an XES file, event by event	–		2.0.1		View
CSV (Event-by-Event)Streams a CSV file, event by event	–		2.0.1		View
Streaming DFG DiscoveryUpdates a DFG with every event			2.0.1		View
Streaming Conformance CheckingChecks every event as it arrives	–	–	2.0.1		View
Token-Based ReplayToken-based replay, event by event			2.0.1		View
FootprintsFootprint checks, event by event			2.0.1		View
Temporal ProfileTemporal profile checks, event by event			2.1.4.1		View
Approximate Online Alignments (IWS)Online alignments with decay time			2.7.23.2	Publication I Will Survive: An Online Conformance Checking Algorithm Using Decay Timeby Raun and Awad	View
OCEL StreamingOne event stream per object type			2.2.16		View

LLM Integration

PM4Py features: LLM Integration
Feature	Input	Output	Added In	Scientific Publication	Source Code	Examples
Textual AbstractionsLogs and models as text for an LLM prompt			2.7.5			View
LLM ConnectorsSends a prompt to an LLM			2.7.5			View

Visual Analytics

PM4Py features: Visual Analytics
Feature	Added In	Scientific Publication	Examples
Dotted ChartEvery event as a dot over time	2.2.7		View
Distribution GraphsDurations, events over time, attributes	2.0.0		View
Performance SpectrumDurations between steps, case by case	2.2.7	Publication The performance spectrum miner: Visual analytics for fine-grained performance analysis of processesby Denisov et al.	View

PM4Py — Process Mining for Python Algorithms and Features

Process Model Types

Event Data

Process Discovery

Conformance Checking

Model Evaluation

Process Enhancement and Analysis

Organizational Mining

Process Simulation

Machine Learning

Streaming Process Mining

LLM Integration

Visual Analytics