Explain how you would approach a take-home case study that includes data prep, defining metrics, building and evaluating a model, and presenting assumptions to management.
Use the provided observations, predictions, and assumptions data to demonstrate the SQL portion of that approach. Include only accepted, sufficiently complete observations and the latest prediction for each record and model version.
Output
- One row per model version and segment with the valid record count, mean absolute error, root mean squared error, model rank, and prior-model error.
- Include the total assumption count and high-confidence assumption count on every row.
- Order by segment ascending, then model version descending.
