Build a production-ready regression model on tabular data using XGBoost with conformalized quantile regression for prediction intervals. Use when the user needs to predict a continuous target from tabular features (price, sales, demand, time-to-event, score) and report uncertainty alongside the point estimate. Default to this for any tabular regression task.
For tabular regression, default to XGBoost as the point estimator and use conformalized quantile regression (CQR) to attach prediction intervals that actually achieve their stated coverage. Point estimates without intervals are not a model — they're a guess. This skill teaches the workflow for shipping a regressor that a downstream system can trust.
Ridge, Lasso, ElasticNet) will be hard to beat and gives
closed-form coefficient interpretation<project>/
├── data/ # input parquet/csv
├── src/
│ ├── train.py # ibis read → 3 XGBRegressors → conformal cal → MLflow
│ ├── predict.py # reload models + conformal_q, return point + interval
│ └── plots.py # predicted vs actual, residual diagnostics, coverage, SHAP
├── notebooks/
│ └── demo.py # marimo walkthrough
└── mlruns/ # MLflow tracking store (gitignored)
Use ibis (ibis-framework[duckdb]) to read data and compute
summaries; materialize with .execute() exactly once just before