Media Bias BR (NLP)
PythonNLPClustering
Scraped ~600 news articles; measured sentiment/subjectivity toward candidates.
About
Sobre
I’m a data analyst with a Master’s from FGV/EBAPE. I run end-to-end projects—scraping, integration, statistical/causal modeling, NLP, and executive reporting. My toolkit spans regressions, DiD, RDD, CEM, plus NLP (classification, summarization, sentiment, topic clustering) for policy and sector analyses.
At KPMG (UNDP & INSS), I led nationwide field interviews, merged qualitative insights with administrative/public data, built personas, and delivered an analytical report with diagnostics, stakeholder mapping, a comms plan, and operational recommendations. Deliverables and scripts, however, are confidential and cannot be included in this portfolio.
This portfolio compiles author-driven work from my MBA/Master’s. In my dissertation (PNAB vs. Rouanet), I built administrative-capacity indices, political alignment/ideology indicators, and estimated effects using CEM with SATT/OLS and bootstrap. I also built NLP studies on media bias/HHI, a systematic review on Green Supply Chain & Digital Transformation, and lab replications of DiD (Ceará) and RDD (electronic voting).
Featured Projects
Projetos em Destaque
Bibliometrics: Green Supply Chain × Digital
RBibliometrixText Mining
State-of-the-art mapping with Bibliometrix; thematic clusters (economic, socio-environmental, internal-external).
Lei Rouanet: Cultural Funding Analysis
PythonAPIRegression
SALIC + IBGE integration; project counts × city traits; evidence of size/income disparities and distance effects.
PNAB: Municipal Determinants of Federal Cultural Funds
RCEMPolicy
Master's dissertation project on determinants of municipal access/implementation of federal cultural funds (PNAB). Integrates TSE, IBGE/MUNIC, FINBRA, SICONFI, ideology datasets; builds capacity indices; uses CEM + OLS/bootstrapping.
Skills & Tools
Habilidades & Ferramentas
Python (pandas, numpy, scikit-learn), R (fixest, Bibliometrix), NLP (TextBlob, tidytext), Causal inference (DiD, RDD), Web scraping (newspaper3k), Data viz (ggplot2, matplotlib), Reproducibility (Git/GitHub, RMarkdown, notebooks).