Publications

@article{RING2025_Martin,

title = {Metalearning for improving time series forecasting based on deep learning: A water case study},

author = {Martín Solís and A. Gil-Gamboa and Alicia Troncoso},

url = {https://www.sciencedirect.com/science/article/pii/S2590123025035960},

doi = {10.1016/j.rineng.2025.107541},

year  = {2025},

date = {2025-12-09},

urldate = {2025-12-09},

journal = {Results in Engineering},

volume = {28},

pages = {107541},

keywords = {deep learning, forecasting, time series},

pubstate = {published},

tppubtype = {article}

}

@article{GIL-GAMBOA25,

title = {Energy-efficient transfer learning for water consumption forecasting},

author = {A. Gil-Gamboa and J. F. Torres and F. Martínez-Álvarez and A. Troncoso},

url = {https://www.sciencedirect.com/science/article/pii/S2210537925000502},

doi = {https://doi.org/10.1016/j.suscom.2025.101130},

year  = {2025},

date = {2025-05-07},

urldate = {2025-05-07},

journal = {Sustainable Computing: Informatics and Systems},

volume = {46},

pages = {101130},

abstract = {Artificial intelligence is expanding at an unprecedented rate due to the numerous advantages it provides to all types of businesses and industries. Water utilities are adopting artificial intelligence models to optimize water management in cities nowadays. However, the substantial computational demands of artificial intelligence present challenges, particularly regarding energy consumption and environmental impact. This paper addresses this problem by proposing a transfer learning approach for water consumption forecasting that reduces computational time, energy usage, and CO$_2$ emissions. The proposed methodology consists in developing a transfer learning approach based on a deep learning model already trained for a task with similar characteristics such as predicting electricity consumption. Thus, a pre-trained deep learning model designed for electricity consumption prediction is adapted to the water consumption domain, leveraging shared characteristics between these tasks. Experiments are conducted to determine the optimal amount of knowledge transfer and compare the performance of this approach with other state-of-the-art time-series forecasting models. Using real data from a water company in Spain, the transfer learning model achieves a similar or better accuracy than the other methods, while demonstrating significantly lower computational times, energy consumption and CO2 emissions. In addition, a scalability analysis has been conducted leading to the conclusion that the proposed transfer learning model is highly suitable to deal with big data. These findings highlight the potential of transfer learning as a sustainable and scalable solution for big data challenges in water management systems.},

keywords = {deep learning, forecasting, time series, transfer learning},

pubstate = {published},

tppubtype = {article}

}

@article{TRONCOSO-GARCIA25,

title = {A new metric based on association rules to assess explainability techniques for time series forecasting},

author = {A. R. Troncoso-García and M. Martínez-Ballesteros and F. Martínez-Álvarez and A. Troncoso},

url = {https://ieeexplore.ieee.org/document/10879535},

doi = {10.1109/TPAMI.2025.3540513},

year  = {2025},

date = {2025-02-11},

urldate = {2025-02-11},

journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence},

volume = {47},

number = {5},

pages = {4140-4155},

abstract = {This paper introduces a new, model-independent, metric, called RExQUAL, for quantifying the quality of explanations provided by attribution-based explainable artificial intelligence techniques and compare them. The underlying idea is based on feature attribution, using a subset of the ranking of the attributes highlighted by a model-agnostic explainable method in a forecasting task. Then, association rules are generated using these key attributes as input data. Novel metrics, including global support and confidence, are proposed to assess the joint quality of generated rules. Finally, the quality of the explanations is calculated based on a wise and comprehensive combination of the association rules global metrics. The proposed method integrates local explanations through attribution-based approaches for evaluation and feature selection with global explanations for the entire dataset. This paper rigorously evaluates the new metric by comparing three explainability techniques: the widely used SHAP and LIME, and the novel methodology RULEx. The experimental design includes predicting time series of different natures, including univariate and multivariate, through deep learning models. The results underscore the efficacy and versatility of the proposed methodology as a quantitative framework for evaluating and comparing explainable techniques.},

keywords = {association rules, forecasting, time series, XAI},

pubstate = {published},

tppubtype = {article}

}

@article{CASAS-GOMEZ25,

title = {Forecasting basal area increment in forest ecosystems using deep learning: A multi-species analysis in the Himalayas},

author = {P. Casas-Gómez and J. F. Torres and J. C. Linares and A. Troncoso and F. Martínez-Álvarez},

url = {https://www.sciencedirect.com/science/article/pii/S157495412400493X?via%3Dihub},

doi = {https://doi.org/10.1016/j.ecoinf.2024.102951},

year  = {2025},

date = {2025-01-01},

journal = {Ecological Informatics},

volume = {85},

pages = {102951},

abstract = {This study addresses the task of forecasting Basal Area Increment trends in forest ecosystems, which is essential for conservation and biodiversity management, particularly in the context of climate change. Traditional forecasting techniques, such as Linear Mixed Models, Random Forest and standard Artificial Neural Networks, often fail to account for the time-dependent nature of tree growth and utilize simple architectures. To overcome these limitations, we introduce the use of two different Deep Learning models: the Long Short-Term Memory network and the Temporal Convolutional Neural Network, which capture the temporal dependencies of growth by incorporating lagged Basal Area Increment values. Our methodology includes rigorous hyperparameter tuning to optimize the Deep Learning models’ architecture. We evaluate the models’ performance across 15 species in the Himalayan region, individually and collectively, using temperature and precipitation data as predictors. The Deep Learning model significantly outperforms state-of-the-art techniques, achieving the lowest Root Mean Squared Error (7.407 for LSTM and 6.202 for TCNN), highest 𝑅2 (0.495 for LSTM an0.585 for TCNN) and lowest Mean Absolute Percentage Error values (37.653 for LSTM and 34.296 for TCNN). These findings highlight the potential of Deep Learning networks to provide accurate and reliable Basal AreaIncrement forecasts, offering valuable insights for forest management and conservation efforts in the face of ongoing climate change.},

keywords = {deep learning, time series},

pubstate = {published},

tppubtype = {article}

}

@article{Melgar2025,

title = {Online forecasting using neighbor-based incremental learning for electricity markets},

author = {L. Melgar-García and D. Gutiérrez-Avilés and C. Rubio-Escudero and A. Troncoso },

url = {https://link.springer.com/article/10.1007/s00521-024-10876-x},

doi = {https://doi.org/10.1007/s00521-024-10876-x},

year  = {2025},

date = {2025-01-01},

urldate = {2025-01-01},

journal = {Neural Computing and Applications},

keywords = {energy, IoT, time series},

pubstate = {published},

tppubtype = {article}

}

@article{Vellinger2025Livestock,

title = {Forecasting Livestock Activity through Interpretable Neuroevolutionary Transfer Learning},

author = {A. Vellinger and F. Rodriguez-Diaz and F. Divina and J. F. Torres},

url = {https://doi.org/10.1093/jigpal/jzaf034},

doi = {10.1093/jigpal/jzaf034},

year  = {2025},

date = {2025-01-01},

journal = {Logic Journal of the IGPL},

volume = {to appear},

number = {to appear},

pages = {to appear},

abstract = {In this paper, we describe a neuroevolutionary approach to livestock activity forecasting, specifically targeting the prediction of Iberian pigs movements. We successfully integrated Transfer Learning to save computational time and used an Explainable Artificial Intelligence technique to provide valuable insights from the model predictions. Inspired by previous work, we employ Deep Evolutionary Network Structured Representation to optimize both Long Short-Term Memory networks and Convolutional Neural Networks using genetic algorithms and dynamic structured grammatical evolution, and we compare the results with other commonly used approaches for time series forecasting. Experimental results demonstrate the superior performance of the proposed Long Short-Term Memory models over more traditional methods, highlighting their precision and consistency in predicting livestock activities. Furthermore, the application of Explainable Artificial Intelligence techniques enable to gain a deeper understanding and trust in AI-driven decisions within precision livestock farming.},

keywords = {deep learning, pattern recognition, time series},

pubstate = {published},

tppubtype = {article}

}

@article{RODRIGUEZ24,

title = {Explainable Olive grove and Grapevine pest forecasting through machine learning-based classification and regression},

author = {F. Rodríguez-Díaz and A. M. Chacón-Maldonado and A. R. Troncoso-García and G. Asencio-Cortés},

url = {https://www.sciencedirect.com/science/article/pii/S2590123024013136},

doi = {https://doi.org/10.1016/j.rineng.2024.103058},

year  = {2024},

date = {2024-09-09},

urldate = {2024-09-09},

journal = {Results in Engineering},

volume = {24},

pages = {103058},

abstract = {Pests significantly impact agricultural productivity, making early detection crucial for maximizing yields. This paper explores the use of machine learning models to predict olive fly and red spider mite infestations in Andalusia. Four datasets on crop phenology, pest populations, and damage levels were used, with models developed using the Python package H20, which focuses on interpretability through SHAP values and ICE plots. The results showed high precision in predicting pest outbreaks, particularly for the olive fly, with minimal differences between models using feature selection. In the vineyard dataset, the selection of characteristics improved the performance of the model by reducing the MAE and increasing R2. Explainability techniques identified solar radiation and wind direction as key factors in olive fly predictions, while past pest occurrences and wind velocity were influential for red spider mites, providing farmers with actionable insights for timely pest control.},

keywords = {deep learning, feature selection, time series, XAI},

pubstate = {published},

tppubtype = {article}

}

@article{JIMENEZ-NAVARRO24b,

title = {Embedded feature selection for neural networks via learnable drop layer},

author = {M. J. Jiménez-Navarro and M. Martínez-Ballesteros and I. S. Brito and F. Martínez-Álvarez and G. Asencio-Cortés},

url = {https://academic.oup.com/jigpal/advance-article/doi/10.1093/jigpal/jzae062/7689640},

doi = {https://doi.org/10.1093/jigpal/jzae062},

year  = {2024},

date = {2024-07-06},

urldate = {2024-07-06},

journal = {Logic Journal of the IGPL},

pages = {jzae062},

abstract = {Feature selection is a widely studied technique whose goal is to reduce the dimensionality of the problem by removing irrelevant features. It has multiple benefits, such as improved efficacy, efficiency and interpretability of almost any type of machine learning model. Feature selection techniques may be divided into three main categories, depending on the process used to remove the features known as Filter, Wrapper and Embedded. Embedded methods are usually the preferred feature selection method that efficiently obtains a selection of the most relevant features of the model. However, not all models support an embedded feature selection that forces the use of a different method, reducing the efficiency and reliability of the selection. Neural networks are an example of a model that does not support embedded feature selection. As neural networks have shown to provide remarkable results in multiple scenarios such as classification and regression, sometimes in an ensemble with a model that includes an embedded feature selection, we attempt to embed a feature selection process with a general-purpose methodology. In this work, we propose a novel general-purpose layer for neural networks that removes the influence of irrelevant features. The Feature-Aware Drop Layer is included at the top of the neural network and trained during the backpropagation process without any additional parameters. Our methodology is tested with 17 datasets for classification and regression tasks, including data from different fields such as Health, Economic and Environment, among others. The results show remarkable improvements compared to three different feature selection approaches, with reliable, efficient and effective results.},

keywords = {deep learning, feature selection, time series},

pubstate = {published},

tppubtype = {article}

}

@article{JIMENEZ-NAVARRO24,

title = {Explaining deep learning models for ozone pollution prediction via embedded feature selection},

author = {M. J. Jiménez-Navarro and M. Martínez-Ballesteros and F. Martínez-Álvarez and G. Asencio-Cortés},

url = {https://www.sciencedirect.com/science/article/pii/S1568494624002783},

doi = {https://doi.org/10.1016/j.asoc.2024.111504},

year  = {2024},

date = {2024-04-04},

journal = {Applied Soft Computing},

volume = {157},

pages = {111504},

abstract = {Ambient air pollution is a pervasive global issue that poses significant health risks. Among pollutants, ozone (O3) is responsible for an estimated 1 to 1.2 million premature deaths yearly. Furthermore, O3 adversely affects climate warming, crop productivity, and more. Its formation occurs when nitrogen oxides and volatile organic compounds react with short-wavelength solar radiation. Consequently, urban areas with high traffic volume and elevated temperatures are particularly prone to elevated O3 levels, which pose a significant health risk to their inhabitants. In response to this problem, many countries have developed web and mobile applications that provide real-time air pollution information using sensor data. However, while these applications offer valuable insight into current pollution levels, predicting future pollutant behavior is crucial for effective planning and mitigation strategies. Therefore, our main objectives are to develop accurate and efficient prediction models and identify the key factors that influence O3 levels. We adopt a time series forecasting approach to address these objectives, which allows us to analyze and predict O3 future behavior. Additionally, we tackle the feature selection problem to identify the most relevant features and periods that contribute to prediction accuracy by introducing a novel method called the Time Selection Layer in Deep Learning models, which significantly improves model performance, reduces complexity, and enhances interpretability. Our study focuses on data collected from five representative areas in Seville, Cordova, and Jaen provinces in Spain, using multiple sensors to capture comprehensive pollution data. We compare the performance of three models: Lasso, Decision Tree, and Deep Learning with and without incorporating the Time Selection Layer. Our results demonstrate that including the Time Selection Layer significantly enhances the effectiveness and interpretability of Deep Learning models, achieving an average effectiveness improvement of 9% across all monitored areas.},

keywords = {deep learning, feature selection, time series, XAI},

pubstate = {published},

tppubtype = {article}

}

@article{PEREZ24,

title = {Pattern sequence-based algorithm for multivariate big data time series forecasting: Application to electricity consumption},

author = {R. Pérez-Chacón and G. Asencio-Cortés and A. Troncoso and F. Martínez-Álvarez},

url = {https://www.sciencedirect.com/science/article/pii/S0167739X23004752},

doi = {https://doi.org/10.1016/j.future.2023.12.021},

year  = {2024},

date = {2024-01-29},

journal = {Future Generation Computer Systems},

volume = {154},

pages = {397-412},

abstract = {Several interrelated variables typically characterize real-world processes, and a time series cannot be predicted without considering the influence that other time series might have on the target time series. This work proposes a novel algorithm to forecast multivariate big data time series. This new general-purpose approach consists first of a previous pattern recognition performed jointly using all time series that form the multivariate time series and then predicts the target time series by searching for similarities between pattern sequences. The proposed algorithm is designed to tackle multivariate time series forecasting problems within the context of big data. In particular, the algorithm has been developed with a distributed nature to enhance its efficiency in analyzing and processing large volumes of data. Moreover, the algorithm is straightforward to use, with only two parameters needing adjustment. Another advantage of the MV-bigPSF algorithm is its ability to perform multi-step forecasting, which is particularly useful in many practical applications. To evaluate the algorithm’s performance, real-world data from Uruguay’s power consumption has been utilized. Specifically, MV-bigPSF has been compared with both univariate and multivariate methods. Regarding the univariate ones, MV-bigPSF improved 12.8% in MAPE compared to the second-best method. Regarding the multivariate comparison, MV-bigPSF improved 44.8% in MAPE with respect to the second most accurate method. Regarding efficiency, the execution time of MV-bigPSF was 1.83 times faster than the second-fastest multivariate method, both in a single-core environment. Therefore, the proposed algorithm can be a valuable tool for practitioners and researchers working in multivariate time series forecasting, particularly in big data applications.},

keywords = {big data, energy, forecasting, time series},

pubstate = {published},

tppubtype = {article}

}

@article{GARCIA-SOTO24,

title = {Water consumption time series forecasting in urban centers using deep neural networks},

author = {C. G. García-Soto and J. F. Torres and M. A. Zamora-Izquierdo and J. Palma and A. Troncoso},

url = {https://link.springer.com/article/10.1007/s13201-023-02072-4},

doi = {https://doi.org/10.1007/s13201-023-02072-4},

year  = {2024},

date = {2024-01-12},

journal = {Applied Water Science},

volume = {14},

pages = {1-14},

abstract = {The time series analysis and prediction techniques are highly valued in many application felds, such as economy, medicine and biology, environmental sciences or meteorology, among others. In the last years, there is a growing interest in the sustainable and optimal management of a resource as scarce as essential: the water. Forecasting techniques for water management can be used for diferent time horizons from the planning of constructions that can respond to long-term needs, to the detection of anomalies in the operation of facilities or the optimization of the operation in the short and medium term. In this paper, a deep neural network is specifcally designed to predict water consumption in the short-term. Results are reported using the time series of water consumption for a year and a half measured with 10-min frequency in the city of Murcia, the seventh largest city in Spain by number of inhabitants. The results are compared with K Nearest Neighbors, Random Forest, Extreme Gradient Boosting, Seasonal Autoregressive Integrated Moving Average and two persistence models as naive methods, showing the proposed deep learning model the most accurate results.},

keywords = {deep learning, forecasting, time series},

pubstate = {published},

tppubtype = {article}

}

@article{JIMENEZ-NAVARRO23a,

title = {From Simple to Complex: A Sequential Method for Enhancing Time Series Forecasting with Deep Learning},

author = {M. J. Jiménez-Navarro and M. Martínez-Ballesteros and F. Mártinez-Álvarez and A. Troncoso and G. Asencio-Cortés},

url = {https://academic.oup.com/jigpal/advance-article/doi/10.1093/jigpal/jzae030/7670726},

doi = {10.1093/jigpal/jzae030},

year  = {2024},

date = {2024-01-01},

urldate = {2024-01-01},

journal = {Logic Journal of the IGPL},

volume = {32},

number = {6},

pages = {986-1003},

abstract = {Time series forecasting is a well-known deep learning application field in which previous data are used to predict the future behavior of the series. Recently, several deep learning approaches have been proposed in which several nonlinear functions are applied to the input to obtain the output. In this paper, we introduce a novel method to improve the performance of deep learning models in time series forecasting. This method divides the model into hierarchies or levels from simpler to more complex ones. Simpler levels handle smoothed versions of the input, whereas the most complex level processes the original time series. This method follows the human learning process where general/simpler tasks are performed first, and afterward, more precise/harder ones are accomplished.Our proposed methodology has been applied to the LSTM architecture, showing remarkable performance in various time series. In addition, a comparison is reported including a standard LSTM and novel methods such as DeepAR, Temporal Fusion Transformer (TFT), NBEATS and Echo State Network (ESN).},

keywords = {deep learning, time series},

pubstate = {published},

tppubtype = {article}

}

@article{MARTINEZ24,

title = {Emerging trends in big data analytics and natural disasters (Editorial)},

author = {F. Martínez-Álvarez and R. Scitovski and C. Rubio-Escudero and A. Morales-Esteban},

url = {https://www.sciencedirect.com/science/article/pii/S0098300423001693},

doi = {https://doi.org/10.1016/j.cageo.2023.105465},

year  = {2024},

date = {2024-01-01},

journal = {Computers and Geosciences},

volume = {182},

pages = {105465},

keywords = {big data, natural disasters, time series},

pubstate = {published},

tppubtype = {article}

}

@article{divina2024stacking,

title = {A stacking ensemble learning for Iberian pigs activity prediction: a time series forecasting approach},

author = { F. Divina and M. García-Torres and F. Gómez-Vela and D. S. Rodriguez-Baena},

url = {https://www.aimspress.com/article/doi/10.3934/math.2024652},

doi = {10.3934/math.2024652},

year  = {2024},

date = {2024-01-01},

urldate = {2024-01-01},

journal = {AIMS Mathematics},

volume = {9},

number = {5},

pages = {13358–13384},

abstract = {Automatic determination of abnormal animal activities can 

be helpful for the timely detection of signs of health and welfare 

problems. Usually, this problem is addressed as a classification 

problem, which typically requires manual annotation of behaviors. This 

manual annotation can introduce noise into the data and may not always 

be possible. This motivated us to address the problem as a time-series 

forecasting problem in which the activity of an animal can be predicted. 

In this work, different machine learning techniques were tested to 

obtain activity patterns for Iberian pigs. In particular, we propose a 

novel stacking ensemble learning approach that combines base learners 

with meta-learners to obtain the final predictive model. Results confirm 

the superior performance of the proposed method relative to the other 

tested strategies. We also explored the possibility of using predictive 

models trained on an animal to predict the activity of different animals 

on the same farm. As expected, the predictive performance degrades in 

this case, but it remains acceptable. The proposed method could be 

integrated into a monitoring system that may have the potential to 

transform the way farm animals are monitored, improving their health and 

welfare conditions, for example, by allowing the early detection of a 

possible health problem.},

keywords = {time series},

pubstate = {published},

tppubtype = {article}

}

@article{ESWA2024,

title = {Medium-term water consumption forecasting based on deep neural networks},

author = {A. Gil-Gamboa and P. Paneque and O. Trull and A. Troncoso},

url = {https://www.sciencedirect.com/science/article/pii/S095741742400099X},

doi = {10.1016/j.eswa.2024.123234},

year  = {2024},

date = {2024-01-01},

urldate = {2024-01-01},

journal = {Expert Systems With Applications},

volume = {247},

pages = {123234},

abstract = {Water consumption forecasting is an essential tool for water management, as it allows for efficient planning and allocation of water resources, an undervalued but indispensable resource for all living beings. With the increasing demand for accurate and timely water forecasting, traditional forecasting methods are proving to be insufficient. Deep learning techniques, which have shown remarkable performance in a wide range of applications, offer a promising approach to address the challenges of water consumption forecasting. In this work, the use of deep learning models for medium-term water consumption forecasting of residential areas is explored. A deep feed-forward neural network is developed to predict water consumption of a company’s customers for the next quarter. First, customers are grouped according to their consumption as these customers include both household consumers and special consumers such as public swimming pools, sports halls or small industries. Then, a deep feed-forward neural network is designed for household customers by obtaining the optimal values for those hyperparameters that have a great influence on the network performance. Results are reported using a real-world dataset composed of the water consumption from 1999 to 2015 on a quarterly basis, corresponding to 3262 clients of a water supply company. Finally, the proposed algorithm is evaluated by comparing it with other reference algorithms including an LSTM network.},

keywords = {deep learning, forecasting, time series},

pubstate = {published},

tppubtype = {article}

}

@article{Melgar2024,

title = {A novel incremental ensemble learning for real-time explainable forecasting of electricity price},

author = {L. Melgar-García and A. Troncoso},

url = {https://www.sciencedirect.com/science/article/pii/S0950705124012085},

doi = {https://doi.org/10.1016/j.knosys.2024.112574},

year  = {2024},

date = {2024-01-01},

urldate = {2024-01-01},

journal = {Knowledge-Based Systems},

volume = {305},

pages = {112574},

keywords = {energy, IoT, time series, XAI},

pubstate = {published},

tppubtype = {article}

}

@article{HADJOUT23,

title = {Electricity consumption forecasting with outliers handling based on clustering and deep learning with application to the Algerian market},

author = {D. Hadjout and A. Sebaa and J. F. Torres and F. Mártinez-Álvarez},

url = {https://www.sciencedirect.com/science/article/abs/pii/S0957417423006255},

doi = {https://doi.org/10.1016/j.eswa.2023.120123},

year  = {2023},

date = {2023-10-01},

journal = {Expert Systems with Applications},

volume = {227},

pages = {120123},

abstract = {The reduction of electricity loss and the effective management of electricity demand are vital operations for production and distribution electricity enterprises. To achieve these goals, accurate forecasts of aggregate and individual electricity consumers are necessary. A novel multistep forecasting method is developed to forecast medium-term electricity consumption of the Algerian economic sector. The proposed method goes through the following three steps: cleaning steps, clustering steps and forecasting step of each cluster. The aim of the first step is to detect and then replace outliers. To complete the first phase, Robust Exponential and Holt-Winters Smoothing algorithms are adapted. Then, to carry out accurate forecasting at a lowest level, K-Shape and K-Means clustering methods are utilized to extract similarities and identify customer consumption patterns as a second step. The third step entails developing a deep learning model based on Gated Recurrent Units to forecast the electricity consumption in each cluster. To validate the proposed method, we compared our results to the most known methods in literature like Autoregressive Integrated Moving Average, Seasonal Grey Model, LSTM networks, Temporal Convolutional Networks and two ensemble models. The results of several experiments conducted with 2000 electricity consumers during 14 years from an Algeria province (Bejaia) demonstrate that the proposed method provides remarkable prediction performances. Thus, prediction performances of the K-Shape-based clustering method reach much higher prediction accuracy. According to the MAPE metric, the results of the best predictions are equal to 2.04%. It is also notable that 87% of the clients have a considerably low prediction error.},

keywords = {clustering, deep learning, energy, time series},

pubstate = {published},

tppubtype = {article}

}

@article{JIMENEZ-NAVARRO23c,

title = {A New Deep Learning Architecture with Inductive Bias Balance for Oil Temperature Forecasting},

author = {M. J. Jiménez-Navarro and M. Martínez-Ballesteros and F. Martínez-Álvarez and G. Asencio-Cortés},

url = {https://journalofbigdata.springeropen.com/articles/10.1186/s40537-023-00745-0},

doi = {https://doi.org/10.1186/s40537-023-00745-0},

year  = {2023},

date = {2023-05-28},

journal = {Journal of Big Data},

volume = {10},

pages = {80},

abstract = {Ensuring the optimal performance of power transformers is a laborious task in which the insulation system plays a vital role in decreasing their deterioration. The insulation system uses insulating oil to control temperature, as high temperatures can reduce the lifetime of the transformers and lead to expensive maintenance. Deep learning architectures have been demonstrated remarkable results in various fields. However, this improvement often comes at the cost of increased computing resources, which, in turn, increases the carbon footprint and hinders the optimization of architectures. In this study, we introduce a novel deep learning architecture that achieves a comparable efficacy to the best existing architectures in transformer oil temperature forecasting while improving efficiency. Effective forecasting can help prevent high temperatures and monitor the future condition of power transformers, thereby reducing unnecessary waste. To balance the inductive bias in our architecture, we propose the Smooth Residual Block, which divides the original problem into multiple subproblems to obtain different representations of the time series, collaboratively achieving the final forecasting. We applied our architecture to the Electricity Transformer datasets, which obtain transformer insulating oil temperature measures from two transformers in China. The results showed a 13% improvement in MSE and a 57% improvement in performance compared to the best current architectures, to the best of our knowledge. Moreover, we analyzed the architecture behavior to gain an intuitive understanding of the achieved solution.},

keywords = {deep learning, time series},

pubstate = {published},

tppubtype = {article}

}

@article{JIMENEZ-NAVARRO23b,

title = {PHILNet: A Novel Efficient Approach for Time Series Forecasting using Deep Learning},

author = {M. J. Jiménez-Navarro and M. Martínez-Ballesteros and F. Martínez-Álvarez and G. Asencio-Cortés},

url = {https://doi.org/10.1016/j.ins.2023.03.021},

doi = {https://www.sciencedirect.com/science/article/pii/S0020025523003183?via%3Dihub},

year  = {2023},

date = {2023-03-03},

journal = {Information Sciences},

volume = {632},

pages = {815-832},

abstract = {Time series is one of the most common data types in the industry nowadays. Forecasting the future of a time series behavior can be useful in planning ahead, saving time, resources, and helping avoid undesired scenarios. To make the forecasting, historical data is utilized due to the causal nature of the time series. Several deep learning algorithms have been presented in this area, where the input is processed through a series of non-linear functions to produce the output. We present a novel strategy to improve the performance of deep learning models in time series forecasting in terms of efficiency while reaching similar effectiveness. This approach separates the model into levels, starting with the easiest and continuing to the most difficult. The simpler levels deal with smoothed versions of the input, whereas the most sophisticated level deals with the raw data. This strategy seeks to mimic the human learning process, in which basic tasks are completed initially, followed by more precise and sophisticated ones. Our method achieved promising results, obtaining a 35% improvement in mean squared error and a 2.6 time decrease in training time compared with the best models found in a variety of time series.},

keywords = {deep learning, time series},

pubstate = {published},

tppubtype = {article}

}

@article{EJIGU23,

title = {A Bayesian Optimization-Based LSTM Model for Wind Power Forecasting in the Adama District, Ethiopia},

author = {E. T. Habtemariam and K. Kekeba and M. Martínez-Ballesteros and F. Mártinez-Álvarez},

url = {https://www.mdpi.com/1996-1073/16/5/2317},

doi = {https://doi.org/10.3390/en16052317},

year  = {2023},

date = {2023-02-19},

journal = {Energies},

volume = {16},

pages = {2317},

abstract = {Renewable energies such as solar and wind power have become promising sources of energy to address the increase in greenhouse gases caused by the use of fossil fuels and to resolve current energy crises. Integrating wind energy into a large-scale electric grid presents a significant challenge due to the high intermittency and nonlinear behavior of wind power. Accurate wind power forecasting is essential for safe and efficient integration into the grid system. Many prediction models have been developed to predict the uncertain and nonlinear time series of wind power, but most neglect the use of Bayesian optimization to optimize the hyperparameters while training deep learning algorithms. The efficiency of grid search strategies decreases as the number of hyperparameters increases, and computation time complexity becomes an issue. This paper presents a robust and optimized Long-Short Term Memory network for forecasting wind power generation in the day ahead in the context of Ethiopia's renewable energy sector. The proposal uses Bayesian optimization to find the best hyperparameter combination in a reasonable computation time. The results indicate that tuning hyperparameters using this metaheuristic prior to building deep learning models significantly improves the predictive performance of the models. The proposed models were evaluated using MAE, RMSE, and MAPE metrics and outperformed both the baseline models and the optimized Gated Recurrent Unit architecture.},

keywords = {deep learning, time series},

pubstate = {published},

tppubtype = {article}

}

@article{TRONCOSO-GARCIA23,

title = {A new approach based on association rules to add explainability to time series forecasting models},

author = {A. R. Troncoso-García and M. Martínez-Ballesteros and F. Mártinez-Álvarez and A. Troncoso},

url = {https://www.sciencedirect.com/science/article/pii/S1566253523000295},

doi = {10.1016/j.inffus.2023.01.021},

year  = {2023},

date = {2023-01-22},

journal = {Information Fusion},

volume = {94},

pages = {169-180},

abstract = {Machine learning and deep learning have become the most useful and powerful tools in the last years to mine information from large datasets. Despite the successful application to many research fields, it is widely known that some of these solutions based on artificial intelligence are considered black-box models, meaning that most experts find difficult to explain and interpret the models and why they generate such outputs. In this context, explainable artificial intelligence is emerging with the aim of providing black-box models with sufficient interpretability. Thus, models could be easily understood and further applied. This work proposes a novel method to explain black-box models, by using numeric association rules to explain and interpret multi-step time series forecasting models. Thus, a multi-objective algorithm is used to discover quantitative association rules from the target model. Then, visual explanation techniques are applied to make the rules more interpretable. Data from Spanish electricity energy consumption has been used to assess the suitability of the proposal.},

keywords = {association rules, time series, XAI},

pubstate = {published},

tppubtype = {article}

}

@article{jimenez2023,

title = {Streaming big time series forecasting based on nearest similar patterns with application to energy consumption},

author = {P. Jiménez-Herrera and L. Melgar-García and G. Asencio-Cortés and A. Troncoso},

url = {https://academic.oup.com/jigpal/advance-article-abstract/doi/10.1093/jigpal/jzac017/6534493?redirectedFrom=fulltext},

doi = {https://doi.org/10.1093/jigpal/jzac017},

year  = {2023},

date = {2023-01-01},

journal = {Logic Journal of the IGPL},

volume = {31},

number = {2},

pages = {255-270},

abstract = {This work presents a novel approach to forecast streaming big time series based on nearest similar patterns. This approach combines a clustering algorithm with a classifier and the nearest neighbors algorithm. It presents two separate stages: offline and online. The offline phase is for training and finding the best models for clustering, classification and the nearest neighbors algorithm. The online phase is to predict big time series in real time. In the offline phase, data are divided into clusters and a forecasting model based on the nearest neighbors is trained for each cluster. In addition, a classifier is trained using the cluster assignments previously generated by the clustering algorithm. In the online phase, the classifier predicts the cluster label of an instance, and the proper nearest neighbors model according to the predicted cluster label is applied to obtain the final prediction using the similar patterns. The algorithm is able to be updated incrementally for online learning from data streams. Results are reported using electricity consumption with a granularity of 10 minutes for 4-hour-ahead forecasting and compared with well-known online benchmark learners, showing a remarkable improvement in prediction accuracy.},

keywords = {energy, IoT, time series},

pubstate = {published},

tppubtype = {article}

}

@article{Melgar2023b,

title = {Identifying novelties and anomalies for incremental learning in streaming time series forecasting},

author = {L. Melgar-García and D. Gutiérrez-Avilés and C. Rubio-Escudero and A.Troncoso},

url = {https://www.sciencedirect.com/science/article/pii/S0952197623005109},

doi = {https://doi.org/10.1016/j.engappai.2023.106326},

year  = {2023},

date = {2023-01-01},

journal = {Engineering Applications of Artificial Intelligence},

volume = {123},

pages = {106326},

keywords = {energy, IoT, time series},

pubstate = {published},

tppubtype = {article}

}

@article{Melgar2023a,

title = {A novel distributed forecasting method based on information fusion and incremental learning for streaming time series},

author = {L. Melgar-García and D. Gutiérrez-Avilés and C. Rubio-Escudero and A. Troncoso},

url = {https://www.sciencedirect.com/science/article/pii/S1566253523000635},

doi = {https://doi.org/10.1016/j.inffus.2023.02.023},

year  = {2023},

date = {2023-01-01},

journal = {Information Fusion},

volume = {95},

pages = {163-173},

keywords = {energy, IoT, time series},

pubstate = {published},

tppubtype = {article}

}

@article{MOLINA22,

title = {DIAFAN-TL: An instance weighting-based transfer learning algorithm with application to phenology forecasting},

author = {M. Á. Molina and M. J. Jiménez-Navarro and R. Arjona and F. Mártinez-Álvarez and G. Asencio-Cortés},

url = {https://www.sciencedirect.com/science/article/pii/S0950705122008322},

doi = {https://doi.org/10.1016/j.knosys.2022.109644},

year  = {2022},

date = {2022-10-22},

journal = {Knowledge-Based Systems},

volume = {254},

pages = {109644},

abstract = {The agricultural sector has been, and still is, the most important economic sector in many countries. Due to advances in technology, the amount and variety of available data have been increasing over the years. However, compared to other economic sectors, there is not always enough quality data for one particular domain (crops, plantations, plots) to obtain acceptable forecasting results with machine learning algorithms. In this context, transfer learning can help extract knowledge from different but related domains with enough data to transfer it to a target domain with scarce data. This process can overcome forecasting accuracy compared to training models uniquely with data from the target domain. In this work, a novel instance weighting-based transfer learning algorithm is proposed and applied to the phenology forecasting problem. A new metric named DIAFAN is proposed to weight samples from different source domains according to their relationship with the target domain, promoting the diversity of the information and avoiding inconsistent samples. Additionally, a set of validation schemes is specifically designed to ensure fair comparisons in terms of data volume with other benchmark transfer learning algorithms. The proposed algorithm, DIAFAN-TL, is tested with a proposed dataset of 16 plots of olive groves from different places, including information fusion from satellite images, meteorological stations and human field sampling of crop phenology. DIAFAN-TL achieves a remarkable improvement with respect to 15 other well-known transfer learning algorithms and three nontransfer learning scenarios. Finally, several performance analyses according to the different phenological states, prediction horizons and source domains are also performed.},

keywords = {time series, transfer learning},

pubstate = {published},

tppubtype = {article}

}

@article{HADJOUT22,

title = {Electricity consumption forecasting based on ensemble deep learning with application to the Algerian market},

author = {D. Hadjout and J. F. Torres and A. Troncoso and A. Sebaa and F. Martínez-Álvarez},

url = {https://www.sciencedirect.com/science/article/pii/S0360544221033090},

doi = {https://doi.org/10.1016/j.energy.2021.123060},

year  = {2022},

date = {2022-03-15},

journal = {Energy},

volume = {243},

pages = {123060},

abstract = {The economic sector is one of the most important pillars of countries. Economic activities of industry are intimately linked with the ability to meet their needs for electricity. Therefore, electricity forecasting is a very important task. It allows for better planning and management of energy resources. Several methods have been proposed to forecast energy consumption. In this work, to predict monthly electricity consumption for the economic sector, we develop a novel approach based on ensemble learning. Our approach combines three models that proved successful in the field, namely: Long Short Term Memory and Gated Recurrent Unit neural networks, and Temporal Convolutional Networks. The experiments have been conducted with almost 2000 clients and 14 years of monthly electricity consumption from Bejaia, Algeria. The results show that the proposed ensemble models achieve better performance than both the company's requirements and the prediction of the traditional individual models. Finally, statistical tests have been carried out to prove that significance of the ensemble models developed.},

keywords = {deep learning, energy, time series},

pubstate = {published},

tppubtype = {article}

}

@article{losada2022,

title = {Automatic Eligibility of Sellers in an Online Marketplace: A Case Study of Amazon Algorithm},

author = {A. Gómez-Losada and G. Asencio-Cortés and N. Duch-Brown},

url = {https://www.mdpi.com/2078-2489/13/2/44},

doi = {10.3390/info13020044},

year  = {2022},

date = {2022-01-01},

journal = {Information},

volume = {13},

number = {44},

pages = {1--16},

abstract = {Purchase processes on Amazon Marketplace begin at the Buy Box, which represents the buy click process through which numerous sellers compete. This study aimed to estimate empirically the relevant seller characteristics that Amazon could consider featuring in the Buy Box. To that end, 22 product categories from Italy’s Amazon web page were studied over a ten-month period, and the sellers were analyzed through their products featured in the Buy Box. Two different experiments were proposed and the results were analyzed using four classification algorithms (a neural network, random forest, support vector machine, and C5.0 decision trees) and a rule-based classification. The first experiment aimed to characterize sellers unspecifically by predicting their change at the Buy Box. The second one aimed to predict which seller would be featured in it. Both experiments revealed that the customer experience and the dynamics of the sellers’ prices were important features of the Buy Box. Additionally, we proposed a set of default features that Amazon could consider when no information about sellers was available. We also proposed the possible existence of a relationship or composition among important features that could be used for sellers to be featured in the Buy Box.},

keywords = {feature selection, time series},

pubstate = {published},

tppubtype = {article}

}

@article{castan2022,

title = {A new hybrid method for predicting univariate and multivariate time series based on pattern forecasting},

author = {M.A. Castán-Lascorz and P. Jiménez-Herrera and A. Troncoso and G. Asencio-Cortés},

url = {https://www.sciencedirect.com/science/article/pii/S0020025521012226?via%3Dihub},

doi = {10.1016/j.ins.2021.12.001},

year  = {2022},

date = {2022-01-01},

journal = {Information Sciences},

volume = {586},

pages = {611--627},

abstract = {Time series forecasting has become indispensable for multiple applications and industrial processes. Currently, a large number of algorithms have been developed to forecast time series, all of which are suitable depending on the characteristics and patterns to be inferred in each case. In this work, a new algorithm is proposed to predict both univariate and multivariate time series based on a combination of clustering, classification and forecasting techniques. The main goal of the proposed algorithm is first to group windows of time series values with similar patterns by applying a clustering process. Then, a specific forecasting model for each pattern is built and training is only conducted with the time windows corresponding to that pattern. The new algorithm has been designed using a flexible framework that allows the model to be generated using any combination of approaches within multiple machine learning techniques. To evaluate the model, several experiments are carried out using different configurations of the clustering, classification and forecasting methods that the model consists of. The results are analyzed and compared to classical prediction models, such as autoregressive, integrated, moving average and Holt-Winters models, to very recent forecasting methods, including deep, long short-term memory neural networks, and to well-known methods in the literature, such as k nearest neighbors, classification and regression trees, as well as random forest.},

keywords = {energy, pattern recognition, time series},

pubstate = {published},

tppubtype = {article}

}

@article{velazquez2022distribution,

title = {Distribution level Electric current consumption and meteorological data set of the East region of Paraguay},

author = {G. Velázquez and F. Morales and M. García-Torres and F. Gómez-Vela and F. Divina and J.L. Vázquez Noguera and F. Daumas-Ladouce and C. Ayala and D. Pinto-Roaand P. Gardel-Sotomayor},

url = {https://www.sciencedirect.com/science/article/pii/S2352340921009744},

doi = {10.1016/j.dib.2021.107699},

year  = {2022},

date = {2022-01-01},

journal = {Data in Brief},

volume = {40},

pages = {107699},

publisher = {Elsevier pubstate = published},

abstract = {This paper presents a data set with information on meteorological data and electricity consumption in the department of Alto Paraná, Paraguay. The meteorological data were registered every three hours at the Aeropuerto Guarani, Department of Alto Paraná, which belongs to the Dirección Nacional de Aeronáutica Civil of Paraguay. The final data consists of a total of 22.445 records of temperature, relative humidity, wind speed and atmospheric pressure. On the other hand, the electrical energy consumption data set contains a total of 1.848.947 records, all of them coming from the one hundred and fifteen feeders located throughout the Alto Paraná region of Paraguay. Electrical energy consumption data was provided by Administración Nacional de Electricidad (ANDE). The analysis of this data can yield insights regarding the energy consumption in the area.},

keywords = {energy, time series},

pubstate = {published},

tppubtype = {article}

}

@article{morales2022analysisb,

title = {Analysis of Electric Energy Consumption Profiles Using a Machine Learning Approach: A Paraguayan Case Study},

author = {F. Morales and M. García-Torres and G. Velázquez and F. Daumas-Ladouce and P. Gardel-Sotomayor and F. Gómez-Vela and F. Divina and J. L. Vázquez Noguera and C. Sauer Ayala and D. Pinto-Roa},

url = {https://www.mdpi.com/2079-9292/11/2/267},

doi = {10.3390/electronics11020267},

year  = {2022},

date = {2022-01-01},

journal = {Electronics},

volume = {11},

number = {2},

pages = {267},

abstract = {Correctly defining and grouping electrical feeders is of great importance for electrical system operators. In this paper, we compare two different clustering techniques, K-means and hierarchical agglomerative clustering, applied to real data from the east region of Paraguay. The raw data were pre-processed, resulting in four data sets, namely, (i) a weekly feeder demand, (ii) a monthly feeder demand, (iii) a statistical feature set extracted from the original data and (iv) a seasonal and daily consumption feature set obtained considering the characteristics of the Paraguayan load curve. Considering the four data sets, two clustering algorithms, two distance metrics and five linkage criteria a total of 36 models with the Silhouette, Davies–Bouldin and Calinski–Harabasz index scores was assessed. The K-means algorithms with the seasonal feature data sets showed the best performance considering the Silhouette, Calinski–Harabasz and Davies–Bouldin validation index scores with a configuration of six clusters.},

keywords = {big data, energy, time series},

pubstate = {published},

tppubtype = {article}

}

@article{BUI22b,

title = {Deformation forecasting of a hydropower dam by hybridizing a Long Short-Term Memory deep learning network with the Coronavirus Optimization Algorithm},

author = {K.-T. T. Bui and J. F. Torres and D. Gutiérrez-Avilés and V. H. Nhu and F. Martínez-Álvarez and D. T. Bui},

url = {https://onlinelibrary.wiley.com/doi/abs/10.1111/mice.12810},

doi = {https://doi.org/10.1111/mice.12810},

year  = {2021},

date = {2021-11-24},

journal = {Computer-Aided Civil and Infrastructure Engineering},

volume = {37},

pages = {1368-1386},

abstract = {The safety operation and management of hydropower dam play a critical role in social-economic development and ensure people's safety in many countries; therefore, modeling and forecasting the hydropower dam's deformations with high accuracy is crucial. This research aims to propose and validate a new model based on deep learning long short-term memory (LSTM) and the coronavirus optimization algorithm (CVOA), named CVOA-LSTM, for forecasting the deformations of the hydropower dam. The second-largest hydropower dam of Vietnam, located in the Hoa Binh province, is focused. Herein, we used the LSTM to establish the deformation model, whereas the CVOA was utilized to optimize the three parameters of the LSTM, the number of hidden layers, the learning rate, and the dropout. The efficacy of the proposed CVOA-LSTM model is assessed by comparing its forecasting performance with state-of-the-art benchmarks, sequential minimal optimization for support vector regression, Gaussian process, M5' model tree, multilayer perceptron neural network, reduced error pruning tree, random tree, random forest, and radial basis function neural network. The result shows that the proposed CVOA-LSTM model has high forecasting capability (R2 = 0.874, root mean square error = 0.34, mean absolute error = 0.23) and outperforms the benchmarks. We conclude that CVOA-LSTM is a new tool that can be considered to forecast the hydropower dam's deformations.},

keywords = {deep learning, time series},

pubstate = {published},

tppubtype = {article}

}