Role of the definition
This layer stores the machine-learning analysis definition that runs on the prepared dataset of the shared Wanalyzer project. The definition determines model selection, input columns and their combination mode, output-column interpretation, the K-means cluster count, and automatic execution after data preparation. Because column relationships are stored as zero-based indexes, mappings must be verified again when column order changes. When automatic model and automatic combination are used together, the process can generate many results because every model is tried for every non-empty input subset.
Growth effect:
When automatic model and automatic combination are selected together, the estimated result count is model count × input-combination count.