Skip to content

Research theme 04

Statistical Methods for Complex Social Data

How can complex longitudinal, categorical, relational, and behavioral data be represented and analyzed statistically?

A diagram of categorical behavioral trajectories over time, each row representing one individual's sequence of states, periodically resolving into a smooth latent curve and regrouping into clusters of individuals with similar patterns.

Categorical trajectories resolving into latent structure

Much of the data generated by social systems does not fit neatly into ordinary numeric time series: a social media user’s daily behavior is better described as a sequence of categorical states than as a continuous measurement. This theme is about building statistical representations that respect that structure instead of forcing it into a shape convenient for standard tools.

With collaborators, I have worked on categorical-valued functional data analysis — representing densely observed categorical longitudinal data through latent-process estimation and functional principal component analysis — and applied it to clustering social media users by their behavioral trajectories. That work is implemented in the open-source R package catfda and published in the Journal of the American Statistical Association.

Projects

Related projects

Categorical-Valued Functional Data Analysis

Many behavioral processes are observed as categorical trajectories rather than ordinary numeric time series. This work develops methods for representing and clustering densely observed categorical longitudinal data and applies them to social-media behavior.

Read the project ↗
A diagram of categorical behavioral trajectories over time, each row representing one individual's sequence of states, periodically resolving into a smooth latent curve and regrouping into clusters of individuals with similar patterns.

Information Diffusion & Influence Pathways

This line of work studies how information spreads through social systems and how influence can be measured across actors and communities. It spans my doctoral research, diffusion models, and DARPA MIPS work on influence pathways using transfer entropy and network-based methods.

Read the project ↗
A diagram of three online communities connected by directed influence pathways of varying strength, with the dominant direction of influence shifting between community pairs over time.

Publications

Related publications

2026

Clustering Social Media Users Using Categorical-Valued Functional Data Analysis

Champon, X., Staicu, A., Weishampel, A., Jayalath, C., Rand, W.

Journal of the American Statistical Association

Develops categorical-valued functional data analysis methods for latent-process estimation and clustering of social media users' behavioral trajectories.

  • Functional Data Analysis
  • Categorical Data
  • Clustering

2024

Comparing Social Media Communities using Functional Data Analysis

Champon, X., Jayalath, C., Rand, W., Jasser, J., Garibay, I., Garibay, O.

57th Annual Hawaii International Conference on System Science (HICSS)

Applies functional data analysis methods to compare behavioral patterns across different social media communities.

  • Functional Data Analysis