Skip to content
View tommaso-accornero's full-sized avatar

Block or report tommaso-accornero

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don't include any personal information such as legal names or email addresses. Markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
tommaso-accornero/README.md

Hi, I'm Tommaso

Computational Social Scientist | Economics + Data
I use R, Python, and SQL to turn messy social data into research that means something — from NLP pipelines on IMF policy documents to climate vulnerability scores for Spanish schools.

Background

  • MSc Computational Social Science @ Universidad Carlos III de Madrid (2026)
  • Research Assistant @ Sciences Po – InclusiveParl Project (SPARQL + web scraping for European parliamentary representation data)
  • MSc Thesis: BERT-based classification of IMF conditionality discourse (1980–2020) — working paper in progress
  • BSc Economics and Business Economics @ Utrecht University (2022)
  • Family connection: AccorneroVini, a winery in Monferrato, Piedmont — where I first got interested in market data

What I'm Working On

  • Classifying 40 years of IMF conditionality using BERT — MSc thesis, SSRN working paper in progress
  • Building SPARQL + scraping pipelines for cross-European parliamentary representation data at Sciences Po
  • Preparing a working paper on degrowth readiness typologies across 254 EU regions

Featured Projects

European Degrowth Typology
PCA + hierarchical and K-means clustering on 254 EU NUTS-2 regions to identify structural typologies of degrowth readiness. Validated externally against CO₂ emissions and life expectancy. R Eurostat ggplot2

Spanish Schools Heat Vulnerability Pipeline
Production-ready R pipeline integrating 4 government APIs (CartoCiudad, Nominatim, Catastro, AEMET) to assess climate heat vulnerability across 999 schools — 261,406 observations. 90% geocoding accuracy. R API Policy Tool

Customer Churn ML — Telco
Supervised ML pipeline on IBM Telco dataset (7,043 records): churn classification with Random Forest, LASSO, and SVM, and total charges regression (R²=0.91). Focus on class imbalance and interpretable business insight. R caret glmnet

IKEA Furniture ML Models
Price prediction (R²=0.74) and discount classification on 3,694 products using LASSO and Random Forest in R. R Machine Learning

Roma Acceptance in the EU — Multilevel Analysis
Multilevel model (lme4) on Eurobarometer 493 data — 27,438 respondents across 28 EU countries. Multiple imputation via mice. Individual and country-level drivers of Roma workplace acceptance. R lme4 Survey Methods

Changing Fortunes — FT Visualization Replication
Full replication and redesign of the Financial Times "Changing Fortunes" slope chart on US median household income inequality (1999–2014). Pew Research data. R ggplot2 Data Visualization

Tech Stack

R Python SQL Git

Methods: Regression · Causal Inference · Machine Learning · NLP / Text Classification · Multilevel Modeling · PCA & Clustering · Survey Analysis · Web Scraping · Data Visualization
Libraries: tidyverse · ggplot2 · lme4 · mice · caret · glmnet · randomForest · xml2 · sf · factoextra
Data Sources: Eurostat · Eurobarometer · IMF MONA · World Bank · Catastro · AEMET · CartoCiudad

Research & Writing

  • 📊 IMF Conditionality Working Paper — BERT classification of ~12,000 IMF conditions across 40 years. Targeting SSRN submission July 2026.
  • 🔬 Sciences Po – InclusiveParl — Web scraping + SPARQL pipelines for European parliamentary representation databases.

Get in Touch

Email: tommaso.accornero@outlook.com
LinkedIn: linkedin.com/in/tommaso-accornero-4084931a9
Location: Madrid, Spain
Looking for: Summer 2026 data analytics internships in Europe

Profile Views

Pinned Loading

  1. churning-customers-ml churning-customers-ml Public

    Supervised ML pipeline on IBM Telco churn data (7,043 records): 9-model benchmark for churn classification and total charges regression. Focus on class imbalance and cost-sensitive thresholds.

    HTML

  2. european-regional-degrowth-typology european-regional-degrowth-typology Public

    PCA + K-means on 254 EU NUTS-2 regions to identify structural typologies of degrowth readiness. Validated against CO₂ emissions and life expectancy. R + Eurostat

  3. spanish-schools-heat-vulnerability-scraper spanish-schools-heat-vulnerability-scraper Public

    Forked from miguelam-css/data_harvesting_project_uc3m

    Production-ready R pipeline integrating 4 government APIs (CartoCiudad, Nominatim, Catastro, AEMET) to assess heat vulnerability across 999 Spanish public schools — 261,406 observations.

  4. roma-multilevel-eu roma-multilevel-eu Public

    Multilevel model (lme4) of attitudes toward Roma in the workplace across 28 EU countries. Eurobarometer 493, N=27,438. Individual and country-level predictors with multiple imputation via mice.

    HTML

  5. changing-fortunes-dataviz changing-fortunes-dataviz Public

    Full replication and redesign of the Financial Times "Changing Fortunes" slope chart on US median household income inequality (1999–2014). Built in R with ggplot2, cowplot, and sf.

    HTML

  6. ikea-furniture-analysis ikea-furniture-analysis Public

    ML analysis of IKEA furniture pricing: price prediction (R²=0.74) and discount classification on 3,694 products using LASSO and Random Forest in R.

    HTML