Statistics have always shapedthe way I see the world.

Alban Gerschheimer, Montpellier
459,773 marine observations around Europe, mapped in R
Confusion matrices: the errors sit between sea and glacier
The data model, drawn before the first SQL query
Distance to port against distance to offshore sites, by phylum
Six models on 24 images they had never seen
443,637 observations sit in the class furthest from ports
Introduction to SQL, DataCamp
Observations per year, 2010–2023, counted in SQL
Text clustering workflow in Orange
Mean distance to port per year, phylum against phylum
Cluster C1: eleven fashion articles

Studying them became a natural choice. I'm in my third year of MIASHS at Université Paul-Valéry Montpellier 3, and looking for a data internship from May 2027.

Download my CV French version

Numbers frommy own work

459,773: marine observations from GBIF, joined to European ports in SQL and studied in R.
0.999: AUC of the best of seven image classifiers we compared in Orange: a logistic regression.
10.4: CPE credits from five DataCamp courses, September to October 2026.
6: programming languages, Python and R first.
4: spoken languages: French, English, Spanish and Korean.

GitHub, the last twelve months

5 public repositories

Loading GitHub activity…

Projects

All repositories

Maritime pressures in Europe

Biodiversity, spatial data2026

Measuring how close marine biodiversity observations sit to ports and offshore infrastructure. The project combines GBIF data, SQL queries, distance classes and map reading: 459,773 observations, 2010 to 2023.

Deep Learning MIASHS

Machine learning, error analysis2025

Comparing text and image classification methods in Orange Data Mining. A hundred news articles clustered by vocabulary, 56 labelled images classified by seven models; most errors fall between sea and glacier.