Rust4PM - Process Mining Documentation

Python

pip install "r4pm[polars]"

Needs Python 3.9 or newer.
Where a prebuilt wheel exists for your platform, nothing else is needed. Otherwise pip builds the package from source, which needs a Rust toolchain and takes a few minutes.
The polars extra adds the DataFrame features: r4pm.item_to_df, r4pm.import_item_from_df and the r4pm.df module. Without it, everything else works.
To match a specific version of these docs, pin it: pip install "r4pm[polars]==0.6.2".
pip install "r4pm-full[polars]" installs the same r4pm package with DuckDB support compiled in, for OCEL 2.0 .duckdb files. It is about 15 MiB larger. Install one of the two, not both.

Command line

cargo install r4pm

This builds the CLI from source, so it needs a Rust toolchain. Run r4pm without arguments to list all commands. The CLI guide explains arguments and output files.

Rust

cargo add process_mining@=0.6.2

The default build covers XES, OCEL 2.0 XML and JSON, the directly-follows graph and Alpha+++ discovery, alignments, OC-DECLARE and the linked OCEL types. Optional features add formats and integrations:

Feature	What it adds
`ocel-sqlite`	OCEL 2.0 SQLite import and export
`ocel-duckdb`	OCEL 2.0 DuckDB import and export, using a system DuckDB library
`ocel-duckdb-bundled`	The same with DuckDB compiled in (slower first build, no system library needed)
`ocel-bundle`	The bundled OCEL 2.0 CSV format (`.ocel.zip`)
`ocel-bundle-parquet`	Parquet storage inside OCEL bundles
`dataframes`	Conversion between event data and polars DataFrames
`graphviz-export`	PNG and SVG export of models (needs graphviz on the PATH)
`token-based-replay`	Token-based replay conformance checking
`log-splitting`	Splitting event logs into parts
`extraction-blueprint`	Extracting OCELs from relational data with an extraction blueprint
`extraction-dbcon`	Reading SQLite, CSV, Parquet and XLSX sources for extraction
`bindings`	The function registry used by Python, the CLI and Propel
`all`	Every optional feature at once, including PostgreSQL and DuckDB extraction sources

Enable features with cargo add process_mining@=0.6.2 --features ocel-sqlite,dataframes, or in Cargo.toml:

[dependencies]
process_mining = { version = "=0.6.2", features = ["ocel-sqlite", "dataframes"] }

The crate is in active development and its API can change in minor versions, so pin the version you build against.

Checking what you have

Python: pip show r4pm (or r4pm-full) prints the installed version. r4pm.__variant__ is "default" or "full", and r4pm.has_feature("ocel-duckdb") tells whether DuckDB support is available.
Rust: cargo tree -e features -i process_mining lists the enabled features.