processlog 1.2.0

Generate realistic synthetic process event logs for testing and developing process mining software

ProcessLog

Generate realistic synthetic event logs for building and testing process-mining software.

ProcessLog turns a small YAML description of a process into a realistic event log — seeded and reproducible, with resource pools, working calendars, probabilistic variants, and custom attributes — and exports it as CSV, NDJSON, Parquet, or IEEE XES. Every format imports straight into PM4Py, no reshaping.

It's the data source for when you're developing a process-mining tool and don't have (or can't share) real logs.

What it is

   process definition          ProcessLog              event log
       (YAML)        ──────▶   generate, seeded   ──▶  CSV · NDJSON · Parquet · XES
                                                        PM4Py-ready · IEEE XES 1849-2023

• Standards-clean output — CSV, NDJSON, Parquet, and XES, all directly importable into PM4Py; XES is IEEE 1849-2023 compliant with proper namespaces.
• Realistic dynamics — resource pools with speed/consistency, working calendars (business hours, weekends, holidays), probabilistic branching into named process variants, and domain attributes (cost, priority, region, …).
• Reproducible — a seed pins the entire run, so regression fixtures stay byte-stable.
• Configured, not coded — one self-documenting YAML file describes the whole process.

Pairs with mining-lib — the logs you generate here render directly in it, an embeddable browser viewer for interactive Directly-Follows Graphs.

Example

A few rows of a generated CSV — the standard XES column names, ready for PM4Py:

case:concept:name,concept:name,time:timestamp,org:resource,lifecycle:transition
case_0001,submitted,2024-01-02 09:00:00-05:00,,complete
case_0001,intake_validation,2024-01-02 10:30:00-05:00,clerk_002,complete
case_0001,review_in_progress,2024-01-03 09:15:00-05:00,reviewer_001,complete
case_0001,approved,2024-01-05 14:20:00-05:00,reviewer_001,complete

Drop that log into mining-lib and you get an interactive process map:

How to use

pip install processlog          # add [xes] for XES export: pip install "processlog[xes]"

Generate a log from a config — the CLI is the quickest path:

processlog generate -c configs/process_config.yaml -n 1000   # 1000 cases, all formats
processlog validate -c configs/process_config.yaml           # check a config
processlog info     -c configs/process_config.yaml           # summarise the process

Or from Python:

from processlog import load_config, generate_log, export_csv

config = load_config("configs/process_config.yaml")
df = generate_log(config, seed=42, num_cases=1000)
export_csv(df, "events.csv")

The full CLI, Python API, config schema, and output format live in docs/USAGE.md.

Develop

git clone https://github.com/crlsrmrlsz/processlog.git
cd processlog
python3 -m venv venv && source venv/bin/activate
pip install -e ".[dev]"
pytest

License

MIT — see LICENSE.