Skip to content

MC Pipeline Tab

The MC Pipeline tab visually constructs repeated experiments comparing a solved reference model against a solved DGP model.

Building a Pipeline

Add steps from the palette by selecting or dragging them onto the canvas. Select a node to edit its name and parameters in the Step Inspector.

The graph represents data dependencies:

  • The data source node is the single root. It may be a DGP simulation or raw data node.
  • Filters connect directly from the data source and may feed multiple dependent steps.
  • Built-in transforms and custom transforms may consume data, filter outputs, or upstream payloads. They can feed other transforms, tests, or regressions.
  • Tests and regressions are terminal and cannot feed downstream nodes.

Step order does not affect the current built-in test and regression results. The backend infers a valid execution order from the dependency graph.

Target vectors accept comma- or space-separated values. Target matrices use one row per line, with values separated by commas or spaces.

Custom transform nodes use a restricted numpy-oriented function contract so they can be validated and, when bundled, restored from the .sdsge archive.

Running and Inspecting Results

Set the replication count and fail-fast behavior, then select Validate or Run pipeline.

The Run Summary contains three tabs:

Tab Description
Performance Overall and per-step timings, generated-data summaries, and failures.
Tests Test statistics, p-values, confidence intervals, rejection rates, and trace distributions.
Regressions Coefficient distributions, status counts, fit metrics, and OLS diagnostics where available.

Persistence and Clearing

The pipeline graph, node positions, run settings, and latest completed result persist across browser refreshes. Graph edits replace the saved graph without removing the latest result. A successful run replaces the saved result.

The trash button in the Pipeline panel resets the graph to its initial state. Clear workspace clears both the graph and latest persisted result.

Local Persistence

MC workspace state is stored in the browser used to access the localhost GUI. It is not written into the model configuration.