A medida da desigualdadecomo a Ciência de Dados pode auxiliar a busca da isonomia nas transações tributárias individuais

Resumo

This dissertation investigates how data science and machine learning can assist in ensuring the principle of isonomy in individual tax settlements within the Brazilian federal scope. The central problem lies in the difficulty of comparing hundreds of agreements with customized clauses and complex socioeconomic contexts, which creates risks of discriminatory treatment and administrative opacity. The research adopts an exploratory and quantitative method, based on the transposition of legal information into computational models. It is demonstrated that the linear regression technique, associated with Principal Component Analysis (PCA), allows for calculating the average behavior of precedents and establishing the "measure of inequality" — an objective metric of the degree of deviation of a new proposal in relation to the set of previously signed transactions. The study concludes that the use of statistical tools provides verifiability and transparency to the actions of the Office of the Attorney General of the National Treasury (PGFN), allowing decisions to be grounded scientifically and reducing the subjectivism inherent in negotiations. Finally, requirements are proposed for the implementation of a technological platform that integrates these models into the daily workflow.

Ficha do documento

Tipo

Dissertação

Ano

2026

Instituição

Fundação Getulio Vargas

Fonte

Repositório da FGV

Idioma

Português

Acesso

Acesso aberto

Identificador

oai:repositorio.fgv.br:10438/38768

Temas

Finanças PúblicasTecnologiaDadosGovernança

Palavras-chave

Transação tributária individualIsonomiaCiência de DadosRegressão linearMachine learningIndividual tax settlementIsonomyData ScienceLinear regressionDireitoDireito tributário - BrasilImpostos - ArrecadaçãoProcessamento eletrônico de dadosAprendizado do computadorTecnologia e direito