How-Tos

Task-focused guides for the most common things you’ll do in Corridor, grouped by where they fall in the analytical lifecycle. Pick whichever topic matches what you’re trying to do.

Topic	Description
Registration
Register a Table	Bring source data into the platform by registering it as a DataTable. Tables are the foundation that data elements, features, and models read from.
Common Registration Info	The fields shared by every registered object: Name, Alias, Entity, and the Properties section (Description, Permissible Purpose, Group, Keywords).
Register a Data Element	Expose a single column from a registered table as a governed, reusable input.
Register a Data Aggregate	Compute a value across multiple rows (an aggregated Data Element) in Python, Pandas, or Spark.
Register a Feature	Build a derived variable on top of data elements or other features. Features are inputs to models and policies.
Register a Feature Aggregate	Roll up rows from a lower entity level into a single value at the feature's entity (an aggregated Feature).
Register a Model Definition	Import or upload a trained model so it can be governed, simulated, and used inside policies.
Register a Model Explainer	Attach explanation logic to a model: feature-importance contributions or reason codes per prediction.
Register a Policy Decision Workflow	Define decision logic that combines features and models to approve, decline, price, or trigger downstream actions.
Register Reports	Define reusable analytics and visualizations that run on every job, from confusion matrices to tracked metrics over time.
Register a Global Function	Register reusable Python that can be called by alias across aggregates, features, reports, and policies.
Analytics
Run a Simulation	Execute a registered object on a dataset to validate performance and generate evidence for approval.
Understand Job Output	The columns a job's result table produces for each object type (id, inputs, output, transforms, dependent, block/segment/rule), and where you read them.
Compare Two Simulations	Put two simulation results side by side for champion-vs-challenger, version-vs-version, or sample-vs-sample analysis.
Do a Quick Test	Run an object's logic on one record interactively to sanity-check it while iterating.
Run a Single Record	Trace one application through every block, rule, and offer to debug a specific decision.
Run a What-If Analysis	Create a sandbox variant of a policy, change its rules, and simulate the impact without touching production.
Run a Portfolio Analysis	Run policies and features across a slice of your portfolio and assemble the results into a single dataset.
Use the Notebook (Corridor Python Package)	Query, declare, and run registered objects from code in the hosted Notebook using the corridor Python package.
Move to Production
Approve an Object	Move a draft object through the approval workflow so it can be used outside your draft workspace.
What's in the Artifact	Understand the production-ready package the platform produces for an approved object, and what it bundles.
Monitoring
Monitor your Model or Policy	Track performance, drift, and decision distributions on objects running in production.
Set up Alerts	Configure notifications for performance thresholds, drift, and operational events.