Generate your own process mining data
Smart Data Forge is the tool for building your own process mining data: tables, keys, timestamps, causal chains, deliberate deviations and their business consequences — ready to run as DDL and INSERT scripts for PostgreSQL, SQL Server, MySQL or Oracle. You open it and get going.
A demo dataset made of random noise proves nothing. “20% of all cases have rework” is not an insight, it is a random number generator with a percent sign. It only gets interesting once the data carries a known cause: group C suppliers deviate more often on critical material — which creates an inspection backlog, which creates an invoice block, which creates a payment after the discount deadline.
That is what you build here: you fix the ground truth first, generate the data around it, and then check whether Noreja finds the cause. That turns a demo into a test — one whose answer you already know.
Runs entirely in your browser. No install, no server, no data leaving your machine.
Ground truth, not randomness
Cause, affected dimension, process effect, error pattern and business outcome are settled before the first row is generated.
Objects, not a flat log
Order, delivery, inspection, invoice, payment — with real 1:N and N:1 relationships, branches and joins instead of one event column.
Reproducible by seed
A fixed seed returns the same dataset again. Demos run identically every time and tests stay comparable.
A verifiable result
Because the cause is known, you can measure whether an analysis actually finds it — instead of just showing that something looks odd.
What the generator does
Seven steps, from the first table to the collective run. Each one is its own area in the tool.
01 · Define tables
Business objects, dimensions and histories: primary key, foreign keys, primary timestamp, additional timestamps, properties and status columns.
02 · Generate DDL
The matching database schema for PostgreSQL, SQL Server, MySQL or Oracle — optionally as DROP + CREATE for test environments.
03 · Causal chain
The process chain: order, time gaps and distributions, cardinalities, AND/XOR logic, time trends, case count, period and seed.
04 · Special behaviour
Deliberate deviations: overjump, wrong order, rework and abort — conditional too, so a property can drive timing, likelihood or attributes.
05 · Column values
Values and distributions for supplier group, material group, plant, amount, quantity, region or risk class — every property with an analytical purpose.
06 · Generate data
Small first to check, then full: INSERT scripts to download and load into your target database.
07 · Bundling / batching
In expert mode: fixed inspection slots, capacity, backlog, collective runs and how one delay propagates all the way into lost cash discount.
Three steps to a dataset
Leave your email
The generator opens right here on the page — full size, ready to use. If you would rather work offline, download it as well and start it locally.
Hand the file to your AI
Upload the tool to ChatGPT, Claude or Gemini as well. The AI then sees the fields your version actually has and walks you from the customer problem through hypothesis and ground truth to the configuration, step by step.
Generate, load, verify
Generate the data, load it into your database, analyse it in Noreja — and check it against the cause you built in. The configuration saves as JSON, loads back, and can be reviewed by the AI against your ground truth.
At a glance
Open the generator
Enter your email once, then open the tool with one click. Access stays available in this browser.
Unlock Smart Data Forge
Leave your email address — you then open the tool with one click, right here on the page.
We use your address for access and the occasional product update. Unsubscribe at any time. More in our privacy policy.
Frequently asked questions
What is Noreja Smart Data Forge?
Smart Data Forge is a tool for generating your own synthetic process mining data; it runs entirely in the browser. In it you define tables and relationships, a causal process chain, deliberate deviations such as overjump, wrong order, rework or abort, and value distributions for the properties. The tool then produces DDL and INSERT scripts for PostgreSQL, SQL Server, MySQL or Oracle.
Why would I need synthetic process data?
For demos, training, proofs of concept and tests where real customer data must not be used or is not available yet. And because the cause is deliberately built in, you can check whether an analysis actually recovers it — something real data rarely allows, since nobody there knows the truth for certain.
What does ground truth mean here?
Ground truth is the cause deliberately built into the data, together with its effect: the affected dimension, the process step it changes, the resulting error pattern and the business outcome that follows. A good ground truth cannot be read off a ROOT_CAUSE column; it has to be derivable from relationships, timing, properties and comparison groups.
Does anything I enter leave my machine?
No. The tool runs entirely in the browser and makes no network calls: the configuration and the generated scripts stay on your machine, and downloads are created locally. All that is asked for access is an email address.
Which databases are supported?
The DDL export in the provided version supports PostgreSQL, SQL Server, MySQL and Oracle. For test environments there is also a DROP + CREATE mode that replaces existing tables.
How does the AI coach work with the generator?
You upload the tool to an AI such as ChatGPT, Claude or Gemini. That lets the AI read the fields your version actually has, and it then acts as a coach: from the business problem through hypothesis and ground truth to the concrete setting for every single generator step. The saved JSON configuration can afterwards be handed back to the AI to review against your ground truth.
What does access cost?
Nothing. After entering an email address the generator is free to use, and it can also be downloaded and kept locally.
From test data to the real analysis — talk to us
Join companies that are redefining growth and efficiency with intelligent process analysis.