New R package BKP offers advanced statistical modeling for probability surfaces

A new R package named BKP has been developed for estimating input-dependent probability surfaces from various types of response data. This package implements Beta Kernel Process (BKP) and Dirichlet Kernel Process (DKP) models, which utilize kernel-weighted pseudo-count aggregation and beta-binomial conjugacy for closed-form posterior summaries. The BKP package also includes scalable approximations for larger datasets and supports features like evidence borrowing, adaptive priors, and hyperparameter tuning. Applications demonstrated include probability-surface estimation, classification, and real-world data analysis for disease prevalence mapping and species distribution modeling. AI

IMPACT Provides advanced statistical tools for machine learning tasks like classification and probability estimation.

RANK_REASON The cluster describes a new R package for statistical modeling presented in an arXiv paper. [lever_c_demoted from research: ic=1 ai=0.7]

Read on arXiv stat.ML →

• Beta Kernel Process
• Dirichlet Kernel Process
• Jiangyan Zhao
• Loa loa
• R
• TwinBKP
• TwinDKP

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →