César Antônio Ciuffo Moreira

Mr. César Antônio Ciuffo Moreira

Postgraduate Program in Applied Computing, Universidade de Brasília, Brasília, Distrito Federal, Brazil
orcid
PhD Student

Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 January 2024
Verified
Book Chapter
Closed access
RBRenata M. M. Bonecker
Renata M. M. Bonecker
César Antônio Ciuffo MoreiraCésar Antônio Ciuffo Moreira
César Antônio Ciuffo Moreira
DPDayane E.C. Paula
Dayane E.C. Paula
FVFilipe A. N. Verri
Filipe A. N. Verri
Este estudo realiza a identificação do perfil das empresas brasileiras exportadoras, a partir da lista de exportadoras e importadoras divulgada pelo governo brasileiro, dados fornecidos pela SECEX-MDIC (Secretaria de Comércio Exterior do Ministério do Desenvolvimento, Indústria, Comércio e Serviços) e pela Receita Federal do Brasil. Realizada a análise exploratória dos dados, técnicas de análise de componentes principais, de agrupamento (k-médias) e de mineração de regras de associação foram aplicadas. Por meio da análise de componentes principais, considerando dez atributos quantitativos, foi possível identificar uma quantidade de grupos para iniciar o agrupamento via k-médias. Por conseguinte, por meio de k-médias, foi possível validar a quantidade de grupos e analisar as características de cada grupo para os atributos quantitativos selecionados. E, por fim, com a utilização da técnica de mineração de regras de associação, foi possível extrair características interpretáveis de apenas um grupo. Como resultado, há evidências que os dados existentes não são adequados para identificar o perfil das empresas e que, portanto, para atingir o objetivo será necessário enriquecer a base com informações, tais como valor exportado pela empresa, produto exportado e quantidade de países para os quais a empresa exportou, entre outras.


29 January 2026 in Revista Gestão em Análise
Verified
Journal Article
Open access
PDF file
César Antônio Ciuffo MoreiraCésar Antônio Ciuffo Moreira
César Antônio Ciuffo Moreira
This study investigates how unsupervised machine learning techniques can support People Analytics in Brazilian public organizations, a context characterized by low analytical maturity. Using anonymized data from ApexBrasil's active employees between January 2019 and December 2023, the work follows the CRISP-DM framework to compare three clustering methods: K-means, Hierarchical Clustering, and DBSCAN. Although DBSCAN presented higher silhouette indices, it classified a large portion of records as noise, limiting its organizational utility. Thus, K-means with eight clusters was selected as the best balance between technical quality and sample coverage. The resulting clusters were interpreted as distinct workforce profiles, providing an interpretable basis for future People Analytics investigations and workforce heterogeneity assessment. The results demonstrate the feasibility of implementing People Analytics in contexts with limited administrative data, offering a replicable framework for similar public organizations.

19 December 2025 in Revista Gestão & Tecnologia
Verified
Journal Article
Open access
PDF file
César Antônio Ciuffo MoreiraCésar Antônio Ciuffo Moreira
César Antônio Ciuffo Moreira
Objective: The objective of this study is to explore the use of time series projection models in predicting absenteeism within Human Resources (HR). It aims to compare dif- ferent models to improve organizational action planning and decision-making. Methodology/Procedural Methods: This research employs a quantitative approach, using time series models such as ARIMA, Holt-Winters, and SARIMA. These models are applied to predict absenteeism in HR, demonstrating the strengths and limitations of each in the context of organizational planning. Originality/Relevance: This study addresses the gap in the theoretical application of time series projections within the HR field, specifically focusing on absenteeism prediction. It contributes to the academic relevance by providing insights into the practical utility of time series models for HR decision-making, an area that has not been extensively explored. Main Results: The results indicate that time series projection is a powerful tool for improving HR decision-making, offering valuable insights into absenteeism trends and or- ganizational productivity. The comparison of ARIMA, Holt-Winters, and SARIMA models highlights their respective advantages and limitations. Theoretical/Methodological Contributions: This study provides significant contribu- tions by comparing various time series models in the context of absenteeism prediction. It offers a clear understanding of which model performs best in specific HR scenarios and provides metrics for evaluating success. Social/Managerial Contributions: The findings have direct implications for HR man- agers, providing them with actionable tools to predict absenteeism and plan more effec- tively, thereby improving organizational productivity and mitigating potential challenges. Keywords: Time series, absenteeism, ARIMA, Holt-Winters, SARIMA, HR planning