Process Mining

OCEL Event Logs for Process Mining

Generate synthetic event logs with known ground truth for algorithm benchmarking, conformance checking, and process mining research.

The Data Problem in Process Mining

Extracting event logs from ERP systems is the hardest part of any process mining project

Weeks of Extraction

Extracting event logs from SAP (BKPF/EKKO/VBAK tables), Oracle, or Dynamics requires weeks of SQL development, domain expertise, and data cleansing.

Incomplete Timestamps

Production systems often lack granular timestamps. Change documents may be missing, creating gaps in the event sequence that break conformance checking.

Privacy Constraints

Event logs contain employee IDs, customer references, and transaction amounts. Anonymization removes the attributes that make process mining useful.

No Ground Truth

Real-world process logs don't come with labeled variants. Evaluating conformance-checking algorithms requires known reference models that production data can't provide.

8 Process Types

End-to-end business processes with realistic activity sequences and timing distributions

Procure-to-Pay (P2P)

Purchase requisition through invoice payment and GL posting

Order-to-Cash (O2C)

Sales order through delivery, billing, and cash collection

Record-to-Report (R2R)

Journal entry posting through trial balance and close

Hire-to-Retire (H2R)

Employee onboarding through payroll and offboarding

Plan-to-Produce (P2P-M)

Production planning through manufacturing execution

Issue-to-Resolution (I2R)

Incident creation through triage, fix, and closure

Quote-to-Cash (Q2C)

Opportunity through proposal, contract, and revenue recognition

Acquire-to-Dispose (A2D)

Fixed asset acquisition through depreciation and disposal

Built for Process Mining

Variant Analysis

Configurable process variant distributions. Set the happy-path ratio, define deviation patterns, and control rework loop frequencies for realistic process models.

Conformance Checking

Each generated log includes the reference BPMN model. Measure fitness, precision, and generalization against known ground truth — no manual annotation required.

OCEL 2.0 Multi-Object

Native OCEL 2.0 output with multiple object types per event. Track how a purchase order, goods receipt, and invoice interact across the process lifecycle.

pm4py Integration

Generated logs are directly loadable with pm4py. Includes helper scripts for discovery (Alpha, Heuristic, Inductive Miner) and conformance checking out of the box.

Export Formats

Native compatibility with every major process mining platform

XES 2.0

IEEE standard for event log interchange. Compatible with ProM, PM4Py, and academic tooling.

Celonis IBC

Native Celonis Intelligence Base Connector format for direct upload to Celonis EMS.

Disco CSV

Fluxicon Disco-compatible CSV with case ID, activity, and timestamp columns.

Parquet / OCEL 2.0

Columnar format for large-scale analysis. Native OCEL 2.0 JSON-LD for multi-object event logs.

pm4py Integration Example

import vynfi
import pm4py

client = vynfi.Client(api_key="vf_live_...")

job = client.generate(
    sector="manufacturing",
    module="process_mining",
    process_type="procure_to_pay",
    cases=10_000,
    variant_count=25,
    happy_path_ratio=0.65,
    output_format="xes",
)

# Load directly into pm4py
log = pm4py.read_xes(job.download_path())
net, im, fm = pm4py.discover_petri_net_inductive(log)
fitness = pm4py.fitness_token_based_replay(log, net, im, fm)
print(f"Fitness: {fitness['average_trace_fitness']:.3f}")

Live demo

Try it — process mining on VynFi data

This runs pm4py on our Supply Chain OCEL dataset. No setup, no install — just explore.

Skip the extraction — start mining

Generate production-quality event logs in minutes, not weeks. 5,000 free credits to start.

View Pricing