AutoWMM and JAGStree -- R packages for Population Size Estimation on Relational Tree-Structured Data

2025-06-26

AI Summary

Estimating population size for target subgroups in hierarchical tree-structured data poses significant computational complexity, particularly when applying the weighted multiplier method (WMM) across arbitrary tree topologies. Method: This paper introduces an automated framework that (i) enables the first general-purpose implementation of WMM on arbitrary tree structures and (ii) develops JAGStree—a tool that automatically generates JAGS code for hierarchical Bayesian models tailored to input tree topologies. The framework integrates WMM, tree-aware data processing, and MCMC inference to substantially lower modeling barriers and implementation effort. Contributions/Results: (1) Two open-source R packages—AutoWMM and JAGStree—that jointly support end-to-end population size estimation; (2) enhanced analytical reproducibility and scalability; and (3) a standardized, lightweight solution for inferring sparse populations within complex relational networks.

Technology Category

• Machine Learning: Matrix & Tensor Methods
• Reasoning under Uncertainty: Relational Probabilistic Models
• Multiagent Systems: Other Foundations of Multi Agent Systems

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Web Mining and Content Analysis: Models for Web evolution
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems

Abstract

The weighted multiplier method (WMM) is an extension of the traditional method of back-calculation method to estimate the size of a target population, which synthesizes available evidence from multiple subgroups of the target population with known counts and estimated proportions by leveraging the tree-structure inherent to the data. Hierarchical Bayesian models offer an alternative to modeling population size estimation on such a structure, but require non-trivial theoretical and practical knowledge to implement. While the theory underlying the WMM methodology may be more accessible to researchers in diverse fields, a barrier still exists in execution of this method, which requires significant computation. We develop two exttt{R} packages to help facilitate population size estimation on trees using both the WMM and hierarchical Bayesian modeling; extit{AutoWMM} simplifies WMM estimation for any general tree topology, and extit{JAGStree} automates the creation of suitable JAGS MCMC modeling code for these same networks.

Problem

• Estimating population size from tree-structured relational data
• Simplifying complex weighted multiplier method computations
• Automating hierarchical Bayesian modeling for population estimation

Innovation

• R packages for tree-structured population estimation
• AutoWMM simplifies WMM for any tree topology
• JAGStree automates JAGS MCMC modeling code