Research map / Learning theory and algorithms / Prediction and decision losses
Prediction Risk for the Horseshoe Regression
Anindya Bhadra, Jyotishka Datta, Yunfan Li, Nicholas G. Polson, Brandon Willard
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 horseshoe regression instead of lasso/ridge/PCR/adaptive lasso with a sparse-robust signal | Horseshoe regression · Simulation n=100, p=100 to 400 | out-of-sample prediction SSE: When p=100, 200, 300, 400, horseshoe regression has the lowest prediction SSE | Improved |
| horseshoe regression versus lasso at p=500 | Horseshoe regression · Simulation n=100, p=500 | out-of-sample prediction SSE: 160.11 vs 159.56 | No clear effect |
| use SURE to choose the global shrinkage parameter and estimate risk | Horseshoe regression · Simulation n=100 | SURE accuracy | Improved |
| use PCR instead of shrinkage methods | PCR · Simulation n=100, p=500 | prediction SSE: 366.88 vs 160.11 for horseshoe | Hurt |
| add local shrinkage with heavy tails (horseshoe) over ridge global shrinkage | Ridge regression · Simulation n=100, p=500 | prediction SSE: 188.78 -> 160.11 | Improved |
| horseshoe vs lasso on exactly-null components | Horseshoe regression · Simulation component-wise SURE | SURE | No clear effect |
| use horseshoe regression instead of ridge, PCR, lasso, adaptive lasso, MCP, SCAD | Horseshoe regression · NCI-60 pharmacogenomics | prediction MSE: horseshoe regression performed the best in 5 among the total 12 cases | Mixed |
| use horseshoe regression on A3 (vs ridge) | Horseshoe regression · NCI-60 A3 | prediction MSE: 0.77 -> 0.65 | Improved |
| use horseshoe regression on A1, A8, A10 | Horseshoe regression · NCI-60 A1/A8/A10 | prediction MSE: 1.30, 1.34, 1.33 (all methods around or larger than 1) | Hurt |
Closest papers by meaning
- Improved Shrinkage Prediction under a Spiked Covariance Structure (JMLR 2021 · similarity 0.87)
- Scalable Approximate MCMC Algorithms for the Horseshoe Prior (JMLR 2020 · similarity 0.85)
- Stable Regression: On the Power of Optimization over Randomization (JMLR 2020 · similarity 0.84)
- Prediction Under Latent Factor Regression: Adaptive PCR, Interpolating Predictors and Beyond (JMLR 2021 · similarity 0.84)
- Prediction-Powered Adaptive Shrinkage Estimation (ICML 2025 · similarity 0.84)
- Estimating Generalization Performance Along the Trajectory of Proximal SGD in Robust Regression (NeurIPS 2024 · similarity 0.83)
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.
