rdacca.hp

rdacca.hp implements generalized canonical analysis and hierarchical partitioning methods in R to quantify predictor importance and partition variation in multi-response canonical and regression models, with primary application to ecological datasets.

Key Features:

• Generalization of Canonical Analysis: Generalizes canonical analysis to accommodate any number of response variables, handling one slope per response per predictor.
• Quantitative Framework for Predictor Importance: Introduces a quantitative framework that estimates the overall importance of single predictors within multi-response models.
• Variation and Hierarchical Partitioning: Expands variation partitioning and hierarchical partitioning to estimate importance of individual predictors and groups of predictor variables without the previous computational constraint to four predictor matrices.
• Integration of Commonality Analysis: Links commonality analysis with hierarchical partitioning to extend commonality-based interpretation from single-response regression to multiple-response scenarios.

Scientific Applications:

• Ecological Modeling: Applied to ecological datasets to analyze relationships between multiple predictors and multiple response variables.
• Predictor Importance Estimation: Facilitates more accurate estimation of predictor importance across complex multi-response datasets.
• Model Interpretation: Enhances interpretability of models by partitioning variation among predictors and predictor groups.

Methodology:

Demonstrates mathematical linkages between commonality analysis, hierarchical partitioning, and variation partitioning; implements generalization frameworks allowing analysis of any number of responses or predictor variables/groups; implemented in R.