Fernando Alexandrino
Research Subjects
Fernando Alexandrino has a total of 15 co-authors. Here are the top co-authors by number of shared publications:
- Antonio Carlos Silva Mello
Federal Center for Technological Education of Rio de Janeiro (CEFET/RJ)
1 shared publication Antonio Mello
1 shared publicationArthur Lamblet Vaz
1 shared publicationCarla Pacheco
1 shared publicationDiego Carvalho
1 shared publication- Diego Carvalho
Federal Center for Technological Education Celso Suckow da Fonseca, Rio de Janeiro, Brazil
1 shared publication
Latest Publications
- Open Access
- 2 February 2026
- Published by Springer Nature in Computational Economics
A Novel Method for Time Series Prediction with Small Data: Integrating Data Augmentation, Normalization Techniques, and Machine Learning
In recent years, machine learning techniques for time series prediction have gained significant traction due to their effectiveness and versatility. However, these algorithms often require a minimum amount of data to ac hieve optimal performance and avoid overfitting. In many real-world applications, such data is scarce or difficult to obtain, creating challenges for training accurate models. This limitation highlights the importance of developing methods that can work effectively with small datasets. To address this issue, we propose AIDAN, a novel approach that integrates artificial intelligence, data augmentation, and normalization techniques to enhance predictions for small, low-frequency time series. AIDAN employs transformations to diversify data samples while preserving their essential characteristics and explores the impact of normalization to stabilize data for training machine learning algorithms. We evaluated the performance of our approach by varying the prediction horizon across multiple domains, including environmental and socioeconomic data, for the ten largest global economies using heterogeneous datasets. Our findings revealed that AIDAN outperformed both a set of traditional time series prediction methods (ARIMA, ARIMA-GARCH, and Exponential Smoothing) and a simple machine learning ensemble (Jitter + diff + MLP), achieving between 6% and 9% higher accuracy for short-term predictions and approximately 15% for long-term predictions. This research provides an effective framework for advanced machine learning-driven time series prediction in low-frequency and data-scarce scenarios.
- 2 February 2026
- Published by Springer Nature in Computational Economics
Open AccessA Novel Method for Time Series Prediction with Small Data: Integrating Data Augmentation, Normalization Techniques, and Machine Learning
In recent years, machine learning techniques for time series prediction have gained significant traction due to their effectiveness and versatility. However, these algorithms often require a minimum amount of data to ac hieve optimal performance and avoid overfitting. In many real-world applications, such data is scarce or difficult to obtain, creating challenges for training accurate models. This limitation highlights the importance of developing methods that can work effectively with small datasets. To address this issue, we propose AIDAN, a novel approach that integrates artificial intelligence, data augmentation, and normalization techniques to enhance predictions for small, low-frequency time series. AIDAN employs transformations to diversify data samples while preserving their essential characteristics and explores the impact of normalization to stabilize data for training machine learning algorithms. We evaluated the performance of our approach by varying the prediction horizon across multiple domains, including environmental and socioeconomic data, for the ten largest global economies using heterogeneous datasets. Our findings revealed that AIDAN outperformed both a set of traditional time series prediction methods (ARIMA, ARIMA-GARCH, and Exponential Smoothing) and a simple machine learning ensemble (Jitter + diff + MLP), achieving between 6% and 9% higher accuracy for short-term predictions and approximately 15% for long-term predictions. This research provides an effective framework for advanced machine learning-driven time series prediction in low-frequency and data-scarce scenarios.
- Open Access
- 25 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Latin America Transactions
- Vol. 22 (11) , 899-910
D-AI2-M: Ethanol Production Forecasting in Brazil Using Data-Centric Artificial Intelligence Methodology
Ethanol serves as one of Brazils primary biofuels. The country produces two main types of ethanol: i) hydrous ethanol, directly utilized as vehicle fuel, and ii) anhydrous ethanol, presently integrated at a rate of 27% into regular gasoline. In 2023, data from the National Agency of Petroleum, Natural Gas, and Biofuels (ANP) indicated that the total volume of ethanol sold in Brazil (hydrous and anhydrous) was just over 28 million cubic meters (m3), which corresponded to almost 22% of the total volume of liquid fuels sold in the country. These numbers illustrate the importance of this biofuel in Brazil. Just six states account for approximately 90% of Brazilian ethanol production. The logistical challenge arises from production seasonality and the necessity to transport ethanol from production sites to distribution and resale networks. Commonly, such prediction is supported using econometric models, such as ARIMA. Considering the recent advances in Artificial Intelligence, this challenge prompts the research question: Can we enhance monthly hydrous and anhydrous ethanol production prediction for the primary Brazilian-producing states using Artificial Intelligence Models (AIM) How should data be prepared for such an approach This study aims to contribute to logistical planning by employing D-AI2-M - a Data-Centric Artificial Intelligence (DAI) methodology - to aid in selecting AIM for ethanol production time series in the principal Brazilian-producing states. Our quantitative experimental evaluation demonstrates the superior forecasting performance of D-AI2-M in two approaches: i) Local: where different D-AI2-M outperform the benchmark models depending on the specific time series, and ii) Global: where a single D-AI2-M achieves the best mean performance across the complete set of evaluated time series.
- 25 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Latin America Transactions
- Vol. 22 (11) , 899-910
Open AccessD-AI2-M: Ethanol Production Forecasting in Brazil Using Data-Centric Artificial Intelligence Methodology
Ethanol serves as one of Brazils primary biofuels. The country produces two main types of ethanol: i) hydrous ethanol, directly utilized as vehicle fuel, and ii) anhydrous ethanol, presently integrated at a rate of 27% into regular gasoline. In 2023, data from the National Agency of Petroleum, Natural Gas, and Biofuels (ANP) indicated that the total volume of ethanol sold in Brazil (hydrous and anhydrous) was just over 28 million cubic meters (m3), which corresponded to almost 22% of the total volume of liquid fuels sold in the country. These numbers illustrate the importance of this biofuel in Brazil. Just six states account for approximately 90% of Brazilian ethanol production. The logistical challenge arises from production seasonality and the necessity to transport ethanol from production sites to distribution and resale networks. Commonly, such prediction is supported using econometric models, such as ARIMA. Considering the recent advances in Artificial Intelligence, this challenge prompts the research question: Can we enhance monthly hydrous and anhydrous ethanol production prediction for the primary Brazilian-producing states using Artificial Intelligence Models (AIM) How should data be prepared for such an approach This study aims to contribute to logistical planning by employing D-AI2-M - a Data-Centric Artificial Intelligence (DAI) methodology - to aid in selecting AIM for ethanol production time series in the principal Brazilian-producing states. Our quantitative experimental evaluation demonstrates the superior forecasting performance of D-AI2-M in two approaches: i) Local: where different D-AI2-M outperform the benchmark models depending on the specific time series, and ii) Global: where a single D-AI2-M achieves the best mean performance across the complete set of evaluated time series.