sparsenetgls

sparsenetgls combines graph structure learning and generalized least squares regression to improve regression estimation for multivariate normal regression with Gaussian distributed dependent and explanatory variables.

Key Features:

• Graph structure learning integration: Utilizes multiple well-known graph structure learning approaches to estimate the precision matrix.
• Penalized variance-covariance matrix: Uses a penalized variance covariance matrix with a distance tuning parameter of the graph structure in deriving the sandwich estimators in generalized least squares regression.
• Gaussian graphical model assessment: Provides functions for assessing a Gaussian graphical model using the penalized approach.
• ROC curve visualization: Uses Receiver Operative Characteristics curve as a visualization tool in the assessment.

Scientific Applications:

• Multivariate regression analysis: Provides solutions for multivariate regression with Gaussian distributed dependent variables and explanatory variables.
• Gaussian graphical model evaluation: Supports assessment of Gaussian graphical models through penalized approaches with ROC curve-based visualization.

Methodology:

The main function sparsenetgls() estimates the precision matrix using graph structure learning approaches, then derives sandwich estimators in generalized least squares regression using a penalized variance covariance matrix with a distance tuning parameter of the graph structure.