Case-centric event data for an order handling process

Zenodo (CERN European Organization for Nuclear Research)

Abstract

Case-centric event data created for the Summer School on Process Mining in Bolzano, 29 June -3 July 2026. The data describes a simple order handling process. There are 10,000 cases. Each case represents one customer order and records the events occurring while the order is invoiced, paid, and eventually delivered or canceled. There are 63,836 events. The average number of events per order is 6.384. There are 8 unique activities. The eight activities are: place order - A new customer order is created. send invoice - The invoice for the order is sent. send reminder - A payment reminder is sent when payment does not arrive promptly. pay - The customer pays the invoice. prepare delivery - The order is prepared for shipment/delivery. make delivery - The goods are delivered. confirm payment - Receipt of the payment is confirmed. cancel order - The order is canceled instead of being completed normally. Events and cases also have attributes such as price, resource, quantity, and address. There are two timestamps for event (start time and end time). For process discovery, the end time is used. The data is provided in four formats: event_log-10000-orders.csv - A CSV file with one event per row. event_log-10000-orders.xlsx - An XLSX file with one event per row. event_log-10000-orders.xes - An XES file using the IEEE XES standard table-10000-cases.xlsx and table-10000-events.xlsx - Two XLSX files separating case and event attributes. The two files, table-10000-cases.xlsx and table-10000-events.xlsx, can be uploaded to Celonis and declared as separate case and event tables. In table-10000-cases.xlsx, product, price, quantity, and address are case attributes. In table-10000-events.xlsx, resource is an event attribute (next to case, activity, start, and end time). The data is synthetic and created for didactic reasons and to test/illustrate process discovery techniques. The interesting variability occurs around invoicing and payment. An invoice is sent, but payment may occur at different moments relative to the invoice, and some customers require one or more send reminder events before paying. The dataset combines several common process-mining phenomena: optional activities (send reminder), repetition (multiple reminders may be sent), concurrency or different ordering between payment-related activities, and an exceptional outcome (cancel order). The two pdf files show screenshots of analysis results in Celonis and ProM.

Authors 1

1. Wil M. P. van der Aalst corresponding Aachen

   RWTH Aachen University · Fraunhofer Institute for Applied Information Technology

   Affiliation as printed

   RWTH Aachen University

   Celonis

   Fraunhofer Institute for Applied Information Technology