Research Associate with
Teaching Duties in Statistics

Posts by Collection

awards

PhD Studentship

Published:

Studentship offered by the EPSRC for the Centre of Doctoral Training in Statistics and Machine Learning (StatML CDT)

Warner Prize

Published:

Prize awarded for best MSc Statistics research project at Imperial College London.

Maths PhD Symposium 2022 Award

Published:

Runner-up & People’s Choice Award at poster competition at Imperial College London’s annual Maths PhD Symposium (2022).

Maths PhD Symposium 2025 Award

Published:

People’s Choice Award at poster competition at Imperial College London’s annual Maths PhD Symposium (2025).

publications

Benchmarking distance-based partitioning methods for mixed-type data

Published in Advances in Data Analysis and Classification, 2023

Recommended citation: @article{costa2023benchmarking, title={Benchmarking distance-based partitioning methods for mixed-type data}, author={Costa, Efthymios and Papatsouma, Ioanna and Markos, Angelos}, journal={Advances in Data Analysis and Classification}, volume={17}, number={3}, pages={701--724}, year={2023}, publisher={Springer} }
Download Paper

A Deterministic Information Bottleneck Method for Clustering Mixed-Type Data

Published in Under review, 2024

Recommended citation: @misc{costa2024dibmix, title={A Deterministic Information Bottleneck Method for Clustering Mixed-Type Data}, author={Costa, Efthymios and Papatsouma, Ioanna and Markos, Angelos}, year={2024}, eprint={2407.03389}, archivePrefix={arXiv}, primaryClass={stat.ME}, howpublished = {arXiv preprint}, url = {https://arxiv.org/abs/2407.03389}
Download Paper

A Deterministic Information Bottleneck Method for Clustering Mixed-Type Data

Published in Data Science, Classification, and Artificial Intelligence for Modeling Decision Making (IFCS 2024), 2025

Recommended citation: @inproceedings{costa2024deterministic, title={A Deterministic Information Bottleneck Method for Clustering Mixed-Type Data}, author={Costa, Efthymios and Papatsouma, Ioanna and Markos, Angelos}, booktitle={Conference of the International Federation of Classification Societies}, pages={81--88}, year={2024}, organization={Springer} }
Download Paper

Contribution to the discussion of Dümbgen and Davies (2024) ‘‘Connecting Model-Based and Model-Free Approaches to Linear Least Squares Regression’’

Published in Statistica, 2025

Recommended citation: @article{CostaPapatsouma2024, title={Discussion of the Paper "Connecting Model-Based and Model-Free Approaches to Linear Least Squares Regression" by Lutz Dümbgen and Laurie Davies (2024)}, volume={84}, DOI={10.60923/issn.1973-2201/20656}, number={2}, journal={Statistica}, author={Costa, Efthymios and Papatsouma, Ioanna}, year={2025}, pages={107–108} }
Download Paper

A novel framework for quantifying nominal outlyingness

Published in Statistics and Computing, 2025

Recommended citation: @article{costa2025nominal, title={A novel framework for quantifying nominal outlyingness}, author={Costa, Efthymios and Papatsouma, Ioanna}, journal={Statistics and Computing}, volume={36}, number={41}, pages={41--58}, year={2025}, publisher={Springer} }
Download Paper

Sparse clustering via the Deterministic Information Bottleneck algorithm

Published in Submitted to IFCS 2026, 2026

Recommended citation: @misc{costa2026sparsedib, title={Sparse clustering via the Deterministic Information Bottleneck algorithm}, author={Costa, Efthymios and Papatsouma, Ioanna and Markos, Angelos}, year={2026}, eprint={2601.20628}, archivePrefix={arXiv}, primaryClass={stat.ML}, howpublished = {arXiv preprint}, url = {https://arxiv.org/abs/2601.20628}
Download Paper

software

IBclust

Published:

IBclust is an R package for clustering datasets using the Information Bottleneck method and its variants. This package supports datasets with mixed-type variables (nominal, ordinal, and continuous), as well as datasets that are purely continuous or categorical. The IB approach preserves the most relevant information while forming concise and interpretable clusters, guided by principles from information theory.

SONO

Published:

SONO is an R package for computing scores of outlyingness for data sets consisting of nominal variables. It further includes various evaluation metrics for assessing performance of outlier identification algorithms producing scores of outlyingness.

talks

teaching

BSc/MSci Mathematics

Imperial College London, Department of Mathematics

  • Probability and Statistics (Year 1)
  • Probability for Statistics (Year 2)
  • Statistical Modelling I (Year 2)
  • Introduction to Statistical Learning (Years 3 & 4)
  • Mathematics of Business and Economics (Years 3 & 4)
  • Statistical Modelling II (Years 3 & 4)

MSc Statistics

Imperial College London, Department of Mathematics

  • Computational Statistics
  • Data Science
  • Machine Learning