New random forest framework enhances compositional data analysis

Researchers have developed a new random forest framework for analyzing compositional data, which can be represented in three ways: directly on the simplex, in log-ratio coordinates, or as directional data on the unit sphere. The framework uses a log-likelihood gain as its split criterion and introduces two novel distributions: a tangent-normal (TN) distribution for directional data and a multivariate logistic-normal distribution for log-ratio coordinates. These distributions allow for more flexible covariance structures compared to existing methods, and simulations indicate that the TN forest is often the most accurate depending on the data's covariance structure. AI

IMPACT Introduces novel statistical methods that could improve the accuracy and flexibility of machine learning models for compositional data analysis.

RANK_REASON The cluster contains a research paper detailing a new statistical framework for machine learning. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• arXiv
• Dirichlet
• Kent distribution
• machine learning
• multivariate logistic-normal distribution
• SimpleX Chat
• tangent-normal (TN) distribution
• unit sphere

• paper
• model release

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