End-to-End Python implementation of Muço’s (2025) corruption measurement framework. Combines NLP pipeline (regex extraction, Porter stemming, TF-IDF), PCA-based dimensionality reduction, and fixed-effects OLS to quantify institutional quality from Brazilian audit reports. Includes supervised learning robustness checks and LOO sensitivity analysis.
natural-language-processing text-mining text-classification scikit-learn nltk econometrics supervised-learning dimensionality-reduction principal-component-analysis fixed-effects political-economy text-as-data brazilian-data government-transparency portuguese-nlp research-replication corruption-measurement dictionary-based-classification institutional-quality audit-analysis
-
Updated
Dec 14, 2025 - Jupyter Notebook