Data Lineage

What the Lineage tab shows

The Lineage tab is an audit trail of your runs. For each run it shows:

• which stages ran;
• how many rows and columns went into and came out of each stage;
• how long each stage took;
• the configuration the run was started with.

Use it to find where a column was removed, to explain a row count, or to confirm which settings produced a file.

The run list

• Search by run name, file name or session ID.
• Filter by status: All, Completed, Failed or Cancelled. Each filter shows how many runs it matches.
• Summary tiles show total runs, success rate, failed runs and total input rows, across your whole history.
• Export CSV downloads the runs currently visible, with these columns: session_id, name, filename, status, date and rows.

Inside a run

Open a run to see:

• A flow chart of the six core stages: Anomaly, DTC, MDH, CDS, DSM and DSG. Hover over a stage to see its status, duration, shape going in and coming out, and how many columns it added or removed.
• A transformation report table with the same figures for each stage. Export steps downloads it as CSV, with these columns: step, status, duration_seconds, input_rows, output_rows, input_columns, output_columns and column_delta.
• The pipeline configuration: every parameter the run was submitted with.
• The full session ID, with a copy button.

Each stage has one of these statuses:

• Completed.
• Failed.
• Skipped: the stage was switched off, or the run stopped before reaching it.
• Unknown: nothing was recorded for the stage.

Runs from before per-stage tracking was introduced show less detail.

Per-column decisions (Enterprise)

Enterprise runs add a Per-column decisions panel with:

• the model family the data was prepared for;
• a leaderboard, when several families were compared;
• a table giving, for each column and stage, what was applied, whether it was automatic or your override, a plain-language reason, and the alternatives that were considered.

The panel also links to the run's audit report (HTML or JSON) and to its full lineage.

API

Route	Returns
GET /api/v1/sessions	Your runs. Optional filters: q, status, tag, date_from and date_to.
GET /api/v1/sessions/<id>	Run details: processing_parameters and steps. Each step has a name, status, duration_seconds, input and output rows and columns, and a message. Enterprise runs also include an enterprise block.
GET /api/v1/enterprise/lineage/<id>	Full per-column lineage (Enterprise)
GET /api/v1/enterprise/audit-report/<id>	The audit report. Add ?format=json to get JSON (Enterprise).