Automated evaluation
Configure and run a coordinated evaluation across multiple analytical methods and settings.
How automated evaluation works
Automated evaluation creates method sweeps from one dataset without requiring each run to be configured and launched individually.
The workflow can evaluate supported supervised and unsupervised approaches, including analyses in feature or embedded space, according to the selected scope.
Choosing evaluation scope
The evaluation controls let users choose the breadth and depth of the sweep. Ounias selects method-specific presets according to the requested scope.
- Breadth determines which types of analyses and methods are considered.
- Depth governs how extensively the selected approaches are evaluated.
- Method suitability information from dataset profiling can help inform which approaches to consider.
Execution and outputs
Automated evaluation coordinates a three-phase analysis pipeline. Its component runs are retained with their configurations, metrics, and artifacts in the Runs Manager.
The resulting runs can be compared or selected for cross-method consensus.
Interpreting the sweep
An automated sweep explores alternatives; it does not establish which analytical result is scientifically correct. Different approaches can favor different structures or assumptions.
The consolidated Reports product is planned separately from the runs generated by automated evaluation.