Evaluation of Metrics for Assessing Synthetic Tabular Data Quality
Talk presented in the Data Analysis II session, focusing on the evaluation of metrics for assessing synthetic tabular data quality.
Nora Amama-BenHassun
Ph.D. Candidate in Statistics and Operation Research (UPC)
Professional Summary
I am a statistician and PhD candidate in Statistics and Operations Research at the Department of Statistics and Operations Research of the Universitat Politècnica de Catalunya - BarcelonaTech (UPC), and a member of the research group GRBIO. I specializing in the validation of synthetic tabular data.
With a strong academic foundation from the joint UB–UPC program and research conducted within the IDEAI-UPC center in collaboration with Siemens Energy, my work focuses on developing statistical methodologies that ensure data resemblance, utility, and privacy.
I have experience in R and Python development, R Shiny applications, survey analytics, risk modelling, and applied statistical methods across health, biomedical, and industry contexts. I am passionate about creating rigorous, reproducible solutions and applying quantitative methods to real-world challenges.
Education
Ph.D. Candidate in Statistics and Operations Research
Universitat Politècnica de Catalunya (UPC)
Master in Statistics and Operations Research (MESIO)
Universitat Politècnica de Catalunya - Universitat de Barcelona (UPC - UB)
Bachelor’s degree in Statistics
Universitat de Barcelona - Universitat Politècnica de Catalunya (UB-UPC)
Interests
Synthetic Data Generation and Validation
Statistics
Machine Learning and Applied AI
Research Communication
📚 My Research
My research focuses on the validation of synthetic tabular data from a statistical perspective. I work on evaluating and comparing existing validation metrics, establishing guidelines for their correct use, and developing new measures that better capture the relationship between utility and privacy.
A key part of my work involves exploring the full landscape of quality assessments in the synthetic data validation pipeline, including resemblance, utility, privacy, and their inherent trade-offs. My goal is to contribute to a rigorous, well-defined framework that helps researchers and practitioners evaluate synthetic data safely, effectively, and transparently.
👉 Feel free to reach out — I love connecting with researchers and practitioners!
Recent & Upcoming Talks
Statistical validation metrics for synthetic data
Talk presented as part of the GRBIO PhD Students' Talks, offering an overview of metrics used to assess synthetic tabular data quality.
From real to Synthetic Data ensuring quality
Poster presented during the Poster Session and Cocktail, accompanied by a 3-minute talk introducing the main ideas and contributions of the work on evaluating metrics for synthetic …
Recent News
Contact