Research map / Optimization / Nonconvex and smooth optimization
Convex Regression with Interpretable Sharp Partitions
Ashley Petersen, Noah Simon, Daniela Witten
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use CRISP instead of CART for fitting the mean model | CRISP vs CART · Simulation Scenario 4 (n=100) | average variance of fitted values: 0.843 -> 0.0935 | Improved |
| use CRISP instead of CART for fitting the mean model | CRISP vs CART · Simulation Scenario 1 (n=10,000) | average variance of fitted values: 0.111 -> 0.051 | Improved |
| use CRISP instead of CART for fitting the mean model | CRISP vs CART · Simulation Scenario 2 (n=10,000) | average variance of fitted values: 1.42 -> 0.056 | Improved |
| use CRISP instead of CART for fitting the mean model | CRISP vs CART · Simulation Scenario 3 (n=10,000) | average variance of fitted values: 0.692 -> 0.077 | Improved |
| use CRISP instead of CART for fitting the mean model | CRISP vs CART · Simulation Scenario 4 (n=10,000) | average variance of fitted values: 1.89 -> 0.096 | Improved |
| use CRISP instead of TPS | CRISP vs TPS · Simulation Scenario 4 (n=10,000) | average variance of fitted values: 0.061 -> 0.096 | Hurt |
| increase training set size from 100 to 10,000 for CART (ablation) | CART · Simulation Scenario 4 | average variance of fitted values: 0.843 -> 1.89 | Mixed |
| use CRISP instead of CART on housing data (n=100) | CRISP vs CART · California housing (n=100 train) | average prediction variance: CART variance more than three times that of CRISP and TPS | Improved |
| increase training set size on housing data (ablation) | CART vs CRISP/TPS · California housing (100 to 11,198 train) | test MSE gap | Mixed |
| use a smaller lambda in CRISP (ablation) | CRISP · California housing (n=100 train) | scaled test MSE | Improved |
| add additivity constraint to CRISP (FLAM) on non-additive mean models (ablation) | FLAM vs CRISP · Simulation Scenarios 2 and 4 (n=100) | mean squared prediction error | Hurt |
| additivity constraint (FLAM) when true model is additive (ablation) | FLAM vs CRISP · Simulation Scenario 1 (n=100) | mean squared prediction error | Improved |
Closest papers by meaning
- A convex optimization formulation for multivariate regression (NeurIPS 2020 · similarity 0.85)
- Hinge Regression Tree: A Newton Method for Oblique Regression Tree Splitting (ICLR 2026 · similarity 0.84)
- Rank-one Convexification for Sparse Regression (JMLR 2025 · similarity 0.84)
- Efficient Algorithms for Non-convex Isotonic Regression through Submodular Optimization (NeurIPS 2018 · similarity 0.83)
- Robust Contextual Optimization with Missing Covariates (ICML 2026 · similarity 0.83)
- On Volume Minimization in Conformal Regression (ICML 2025 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
