Event logs and the computation switch

Keep a process's log files, preview them, and compute from real data.

The question this screen answers

How does this process actually run? The numbers you type on the Steps tab are design numbers — how long it should take, how much it should cost. An event log holds the real numbers. The Event log screen is where a process keeps its log files and where you pick which one is used for computation.

Open the Event log tab on the process detail page — alongside Scenarios and Simulation. Everything about this process's logs lives in that one tab: upload, preview, turn on/off, check against the design, and the “how the log was read” summary if the process was built from a file.

Three things on the screen

• Upload — each process keeps up to 10 files. Every file is checked on arrival and anything suspicious is listed underneath it.
• Preview — the table is built from the columns in your own file, not from a fixed set of columns. It is the fastest way to see that the upload matched your expectations before you trust the numbers it produces.
• Turn on / off / delete — only one file is used at a time. Turning a new file on turns the previous one off, so there are never two sources of numbers at once.

The “Compute from the event log” switch

With the switch on, each step's time, waiting time, cost and assignee come from the file in use instead of the numbers you typed. Every analysis screen — metric cards, diagram, graph, scenarios, simulation, the social network graph and the .docx report — follows, and the detail page shows an EVENT LOG band right above the metric cards so you always know whose numbers you are reading.

Deliberately NOT taken from the log

• Standard time and standard cost — a standard is a human number: the log cannot know what it should have been, and overlaying it would destroy the very baseline you compare against.
• Structure — steps, flows and branch probabilities stay with the design. Building structure from a log is a different job, and it has its own Rebuild button on each file's row.
• Exported .drawio / .bpmn files — what you download is still your design.

A sample file filled in with your own step ids

The screen offers two sample downloads. Unlike the generic sample log on the landing page, this one already contains the exact step ids of the process you have open — you only replace the data with your real data.

• Excel (.xlsx) — the better choice. The file carries data validation: the MaBuoc cell only accepts real ids (pick from the drop-down), date cells only accept dates, and the cost cell only accepts non-negative numbers. Mistakes are blocked as you type.
• CSV — for anyone exporting data with another tool.

Checks on upload

Files are checked in three layers, and a file that fails any layer cannot be turned on:

1. File shape — missing required columns, no header row, rows whose timestamps cannot be read.
2. Data plausibility — KetThuc before BatDau, negative cost, exact duplicate rows, cases with a single event, ambiguous dd/mm ↔ mm/dd dates.
3. Match against the design — every MaBuoc in the file must be the id of a real step. An unknown id means the numbers land on the wrong step, and a number on the wrong step looks exactly like a correct one.

A file without a MaBuoc column still works: matching then happens by activity name. Steps absent from the log keep their hand-entered numbers, and the screen names them explicitly.