Skip to content
View ATadmuri-hub's full-sized avatar
🎯
Focusing
🎯
Focusing

Block or report ATadmuri-hub

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 is supported. This note will only be visible to you.
Report abuse

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

Report abuse
ATadmuri-hub/README.md

I use data, simulation, and mixed methods to make sense of policy and social questions.

From political science in Istanbul to policy research in Brussels to computational social science in Madrid.


LinkedIn GitHub Email Location


About Me

Computational social scientist trained across three countries: political science and IR in Istanbul, policy research in Brussels, and computational social science in Madrid. I build models that help decide before implementing, from agent-based policy simulators to ML pipelines on 50,000+ observations.

Education MSc Computational Social Science, UC3M (2025-2026) · BA International Studies, UC3M (via Erasmus+ from Yildiz Technical University, Istanbul)
Thesis Agent-Based Modeling of Migrant Integration Through Community Sport · NetLogo + R, 14,110 runs, 23 scenarios · Supervisor: Prof. Anxo Sanchez · Open code and data (Zenodo DOI)
Honors 2x Matricula de Honor, UC3M's highest academic distinction: Statistics & Data Science II · Big Data & AI Ethics
Experience Research Fellow, EuroMed Rights (Brussels, 2025): daily analysis on Syria and Turkiye, policymaker briefings, 12+ NGO coordination
Certifications Bloomberg Terminal: BMC · BFF · ESG · Spreadsheet Analysis (BQL) · McKinsey Forward Program
Languages Arabic (native) · English (C1) · Turkish (C1) · Spanish (B2) · Persian (A2)
Focus areas Agent-based modeling · Causal inference · Machine learning · Text mining · Social network analysis · ESG analytics · Geopolitical risk
Open to Data science · Policy research · Risk and intelligence analysis · ESG analytics · Computational social science

Tech Stack

Languages & Querying

R Python SQL NetLogo

Machine Learning & Statistics

XGBoost Random Forest Elastic Net GAM SHAP PCA Clustering ARIMA Monte Carlo Survival Analysis Causal Inference Econometrics Survey Methods Text Mining

Frameworks & Libraries

tidyverse tidymodels caret scikit-learn pandas ggplot2 igraph

Data Collection & Tools

Web Scraping REST API Git LaTeX Bloomberg Excel Quarto


Featured Projects

MSc thesis: an agent-based policy simulator testing integration-program design before implementation.

  • 14,110 simulation runs across 23 program-design scenarios
  • PRCC global sensitivity analysis
  • Survival analysis (Kaplan-Meier, Cox proportional hazards)
  • Published with open code and data (Zenodo DOI)

NetLogo R ABM 3,800 lines

21 supervised learning models on 50,000+ Airbnb listings across Madrid, Barcelona, Amsterdam.

  • 13 classification + 8 regression models
  • SHAP beeswarm, dependence, and interaction analysis
  • Calibration curves across all model families
  • Counterfactual pathway analysis
  • Kolmogorov-Arnold Networks (KAN) benchmark

R Python SHAP 5,600 lines

12-phase investment analytics framework built on 11,826 Madrid property listings.

  • 9 models: OLS, Fixed Effects, LASSO, Ridge, Elastic Net, Random Forest, XGBoost, GAM, LMM
  • Automated valuation and mispricing detection
  • District-level price forecasts (2026-2027)
  • Portfolio intelligence with configurable investor holdings
  • Motivated-seller classification (AUC 0.62)

R XGBoost sf 3,200 lines

Which countries are most structurally vulnerable to AI-driven economic disruption?

  • PCA on 11 World Bank indicators (200 countries)
  • 3 clustering methods: K-Means, Hierarchical, Gaussian Mixture
  • Composite Vulnerability Index with bootstrap CIs
  • Validated against Oxford AI Readiness Index 2025

R PCA Clustering 2,700 lines

Reproducible harvesting pipeline building a 28-month panel of Spanish political discourse.

  • Wayback CDX API, rvest, httr2 collection pipeline
  • ARIMA forecasting and interrupted time-series analysis
  • Monte Carlo simulation (n=5,000) for D'Hondt seat projection
  • Fully reproducible in R

R Web Scraping ARIMA Team Project

Can small states be big players in development and good governance?

  • Population size vs human development, governance, and democracy
  • World Bank, UNDP, and V-Dem data (1990-2022)
  • Extended analysis of democracy, CO2 emissions, and HDI
  • Fully reproducible R Markdown pipeline

R ggplot2 V--Dem Data Viz

Network structure, link prediction, and spreading on two empirical networks.

  • Copenhagen Bluetooth proximity network (662 nodes)
  • arXiv GR-QC collaboration network (4,158 nodes)
  • Community detection and link prediction
  • SIR epidemic spreading simulation

R igraph SIR Networks

End-to-end ML pipeline: price regression and discount classification on 3,694 IKEA products.

  • 7 classification + 6 regression models
  • Feature engineering on mixed types
  • Cross-validation, RMSE, MAE, ROC-AUC
  • Log-price transformation for skewness correction

R tidymodels glmnet 1,500 lines


What I Deliver

Policy Research & Intelligence Daily analytical reporting, policymaker briefings, EU Parliament coverage, geopolitical risk monitoring (Syria, Turkiye, MENA)
Machine Learning Pipelines End-to-end supervised learning: classification, regression, SHAP interpretability, counterfactual analysis across 50,000+ observations
Agent-Based Modeling Computational simulation of social systems using NetLogo, applied to policy design and evaluation (MSc thesis: 14,110 runs, 23 scenarios)
Causal Inference & Econometrics Counterfactual analysis, survey methodology, social network analysis, bootstrap confidence intervals
ESG & Sustainability Analysis Bloomberg-certified (BMC, BFF, ESG, BQL), quantitative ESG risk assessment, climate risk
Data Collection & Scraping Web scraping (rvest, RSelenium), REST API integration, automated data pipelines for research
Multilingual Analysis Arabic (native), English, Turkish, Spanish: policy briefings, report translation, cross-border NGO coordination (12+ partners)


Open to new opportunities. Let's connect.


LinkedIn


Pinned Loading

  1. Data-Harvesting-uc3m-2026 Data-Harvesting-uc3m-2026 Public

    Reproducible R harvesting pipeline (Wayback CDX, rvest, httr2). 28-month panel, ARIMA forecasts, Monte Carlo (n=5,000) D'Hondt seat projections, interrupted time series.

    HTML 1

  2. Madrid-Housing-Market-Analysis Madrid-Housing-Market-Analysis Public

    12-phase real-estate analytics on 11,826 listings: mispricing detection, district forecasts, motivated-seller classification. Best model XGBoost R2=0.74. R/Quarto.

    R 1

  3. Project-Which-Countries-Are-Most-Vulnerable-to-AI-Driven-Economic-Disruption- Project-Which-Countries-Are-Most-Vulnerable-to-AI-Driven-Economic-Disruption- Public

    PCA + K-Means/hierarchical/GMM clustering with ARI validation, ranking ~200 countries by AI-disruption exposure vs resilience. Validated against Oxford AI Readiness.

    R 1

  4. Airbnb-Superhost-Price-Prediction Airbnb-Superhost-Price-Prediction Public

    21 supervised models (13 classification, 8 regression) on 50,000+ Airbnb listings. SHAP, calibration, bootstrap CIs, counterfactuals. R + Python.

    R 1

  5. CIM-Model CIM-Model Public

    Agent-based model (NetLogo) of migrant integration through community sport. 14,110 runs, 23 scenarios, PRCC sensitivity, survival analysis. MSc thesis. DOI + Zenodo.

    NetLogo

  6. dataviz dataviz Public

    Forked from IBiDat/dataviz

    Homepage and materials for the course on data visualization, as part of uc3m’s Master in Computational Social Science

    HTML