Relational Event Models and Bayesian Penalization

Together with Mahdi Shafiee Kamalabad and Sara van Erp, in my master’s thesis, I am exploring the application of Bayesian penalization techniques to relational event models (REMs), which are used to analyze interaction patterns between actors over time. REMs have become a valuable tool for understanding how various factors influence the rate of interactions in fields ranging from corporate networks to communication systems. However, the process of selecting the most relevant predictors in these models can be challenging due to the complexity of the data, which includes both exogenous variables (like age or gender) and endogenous ones (such as prior interactions between actors).

To address this challenge, I am investigating how Bayesian regularization methods, particularly those using shrinkage priors, can improve variable selection and model performance in REMs. Through simulations and applications to collaborations on Spotify and communications during the Apollo-13 mission, I aim to identify the most effective shrinkage techniques for narrowing down significant predictors and enhancing predictive accuracy. This research will contribute to a deeper understanding of how Bayesian methods can refine relational event modeling, making it easier for researchers to uncover key interaction dynamics.

Data Donation and Digital Trace Data

I am currently involved in data donation through the D3I (Digital Data Donation Infrastructure) project. The project aims to make data donation available to all social science and humanities researchers in the Netherlands by facilitating digital trace data collection from participants, through the EU’s GDPR allowing individuals to download data that platforms gather on them. Funded by PDI-SSH, it lets participants donate personal data, like social media activities or transaction records, through Data Download Packages (DDPs), which are locally processed for privacy. Only necessary data for research, with consent, is shared. D3I provides open-source tools like Port to aid researchers in data donation projects, prioritizing privacy and participant data control. These tools grant researchers access to previously inaccessible digital trace data for studying human behavior.


In addition to that, I am also involved in the preparation of an undergraduate course on Digital Trace Data. In this role, I developed a series of practical exercises focused on a range of topics, including user- and platform-centric approaches as well as the application of AI in analyzing digital trace data using Python. These practicals were designed to deepen students’ understanding of the subject matter and to provide them with hands-on experience. Over the course of seven weeks, bachelor students engaged with these tasks, gradually building their skills in the characteristics of digital trace data and ways in which to analyse them. In this role, I was also involved in the creation of an educational tool for students to easily observe their own digital footprint from ChatGPT, YouTube, Instagram, Netflix, and WhatsApp.

Research Apps and Ecological Momentary Assessment

As a research assistant on a different project, my current focus is on the exploration and implementation of research apps designed for Ecological Momentary Assessment (EMA). This assessment method opens researchers the opportunity to obtain immediate data on participants’ behaviors, emotions, and experiences directly within their natural settings, most commonly through conducting triggered surveys on their smartphones.