The sample task-mining-sample.csv is not another event log of orders. Each row is a UI action by an operator handling an incoming pizza order. Sessions are named like the process cases (PIZZA-TM-001) so you can keep the business in mind, but the grain is clicks, not Case ID / Activity / Timestamp of the process map.
Operators in the file are op_anna, op_boris, op_clara, op_denis, and op_elena. The CSV has 434 UI events in 42 sessions across five applications: CRM, Order Management, Payment ERP, Kitchen Display, Delivery Console.
| App | Screen | Action |
|---|---|---|
| CRM | Incoming Orders | Open order |
| CRM | Customer Profile | Search customer, then copy phone number |
| Order Management | Address Form | Paste address |
| Order Management | Order Details | Review items, update status |
| Payment ERP | Payment Status | Check payment status |
| Kitchen Display | Kitchen Queue | Send to kitchen |
| Delivery Console | Active Deliveries | Assign courier, mark delivered |
The process-map step “Order accepted” looks like one node. On the desk it is a hop across applications, including copy from the customer profile and paste into the address form. That copy/paste is the finding: the same fact exists in two systems, and a person is the integration.
Do not upload this CSV into Process Mining. The columns are timestamp, user_id, session_id, app, screen, element, event_type, value.
task-mining-sample.csv after you map it, not necessarily to the demo.