Nestimate: An R Package for dynamic, probabilistic, and high order network analysis

Area of Science:

• Computational Psychology
• Network Science
• Data Analysis

Background:

• Network estimation is crucial for analyzing complex data from psychometric scales and event data.
• Existing methods for network analysis can be fragmented and lack a standardized workflow.

Purpose of the Study:

• Introduce the Nestimate R package for a unified and standardized approach to network estimation.
• Provide a comprehensive tool for various network models and data types.
• Facilitate robust validation of estimated networks.

Main Methods:

• Nestimate estimates psychological networks using correlation, partial-correlation, graphical lasso, Ising, and mixed graphical models.
• It supports multi-cluster networks, higher-order network models, and dynamic networks (transition, co-occurrence, mixed).
• A unified validation pipeline includes bootstrapping, permutation testing, split-half reliability, and centrality stability analyses.

Main Results:

• The Nestimate package offers a standardized workflow for diverse network estimation tasks.
• It enables the analysis of psychometric, multi-cluster, higher-order, and dynamic networks.
• Integrated validation methods ensure the reliability and stability of network estimates.

Conclusions:

• Nestimate R package provides a versatile and unified platform for psychological network analysis.
• The package supports a wide range of network models and data types, enhancing research capabilities.
• Standardized validation procedures improve the trustworthiness of network findings.