PM4Py

Overview

PM4Py is a process-mining library for Python, aimed at developers, data scientists, and researchers who want to build analysis into code. Its breadth is a strong fit for technical teams working with event logs; proprietary deployments need to weigh the AGPL terms or obtain a commercial license.

Key features

Discover models and check conformance

PM4Py can derive directly-follows graphs, Petri nets, BPMN models, and process trees from event data. That gives analysts several ways to inspect a process without moving their work out of Python. Replay and alignment-based measures compare observed behavior with process models, and root-cause analysis supports investigating deviations. Refresh is manual, so teams planning recurring analysis must arrange their own update workflow.

Handle event logs and multiple object types

The library imports and exports event logs, supports XES, and handles OCEL 1.0 and OCEL 2.0. Object-centric mining can represent events tied to several kinds of objects, such as orders, items, and deliveries. That is useful when a single case-based view would obscure relationships in the data.

Fit process mining into Python workflows

Machine-learning feature extraction for pandas and scikit-learn makes PM4Py relevant to teams combining process analysis with data-science work. Installation is recommended through pip, and Windows and Linux instructions are provided. Visualization adds a dependency: Graphviz must be installed and configured, which is a practical setup step for users who want rendered diagrams.

Pricing

The Community Version costs nothing, billed free of charge. It is intended for teaching, research, and open-source projects under AGPL v3; covered work must be published under AGPL. That makes it a clear fit for qualifying projects, but a poor option for teams that need to keep distributed or network-accessible software closed-source.

The Commercial Version has custom pricing by quote. It permits closed-source commercial use and includes regular releases, early feature access, and dedicated professional support. Commercial evaluation without open-sourcing code is available through a specific evaluation license. This is the relevant route for proprietary deployments, though buyers must request pricing.

Platforms

PM4Py runs on Linux, macOS, and Windows, and is called through an API as a Python library. It is self-hosted rather than a hosted process-mining service. ERP connectors are supported, but users should account for manual data refresh when designing ongoing analysis.

Who it's for

PM4Py suits developers, data scientists, and researchers comfortable building around a Python library, especially those working with XES or object-centric event data and wanting to connect process mining to pandas or scikit-learn. It is less suitable for teams seeking a ready-made hosted interface or automatic refresh. The AGPL obligation is decisive for proprietary use unless the commercial license is acceptable.

Pros and cons

Broad model discovery: directly-follows graphs, Petri nets, BPMN models, and process trees give technical users multiple process representations.

Useful for complex event structures: OCEL 1.0 and 2.0 support and object-centric mining accommodate events linked to multiple object types.

Python data-science fit: feature extraction for pandas and scikit-learn can bring process analysis into existing workflows.

License constraint: AGPL-covered software must be published under AGPL when distributed or made available over a network, which can exclude proprietary deployments.

Setup and upkeep burden: visualization requires configured Graphviz, and data refresh is manual.

Alternatives

Process Mining Software is the broader category directory for comparing process-mining options.

ProcessMind may suit readers wanting a web option with a freemium plan: its free tier allows one member, unlimited models, and 20 AI credits per month.

Mindzie Process Mining is worth considering for a desktop trial-free starting point: its Free Desktop Edition offers full functionality for up to 1,000 cases.

Process.Science may fit readers seeking a paid tool with web and self-hosted platforms and predefined report pages, PPIs, and KPIs.

ProDiscovery is an alternative with cloud plans and an on-premise version available by inquiry.

ABBYY Timeline offers a 30-day trial with unlimited access and supports web and self-hosted deployment.

Cortado is another free option for Linux, macOS, and Windows.

Pega CRM is a paid, web-based alternative with a free trial.

Verdict

Choose PM4Py if your team wants process-mining algorithms inside Python and values its broad discovery, conformance, and object-centric capabilities. Its free version is compelling for teaching, research, and open-source work; proprietary users should proceed only if the commercial license and custom quote fit. Look elsewhere if you need automatic refresh or want to avoid licensing and visualization setup obligations.