This study clusters solar inverters and wind turbines to help Enlitia’s clients optimize resource allocation and operational strategies by identifying similar assets based on historical power production, meteorological data, and power curve characteristics. The project employs Data Mining techniques following the CRISP-DM methodology, emphasizing data cleaning to handle null values, duplicates, and outliers. For wind turbines, outliers are managed using power curves, while for solar inverters, I-V curves are utilized. Clustering begins after data cleaning, using algorithms from classical, ensemble, and time series clustering categories. Principal Component Analysis is applied to reduce computational costs while preserving significant data variation. Classical clustering involves five hierarchical, two partitional, one soft, one model-based, and two density-based algorithms, evaluated using four distinct indices. The top three classical algorithms proceed to ensemble clustering, combining the three algorithms via weighted major voting. Lastly, two time series clustering algorithms are applied to pre-processed datasets. Evaluation of segmentations indicates that time is a significant factor in data variation. Time series clustering consistently produces the best segmentations for both solar and wind datasets.

This paper explores the discrepancies between laboratory and real-world stress detection, emphasizing the pronounced differences in data loss, data preprocessing, feature design, and classifier selection. Laboratory studies offer a controlled environment that optimizes data quality, whereas real-world settings introduce chaotic and unpredictable elements, coupled with a diverse range of human behaviours, resulting in substantial data loss and compromised data quality. We discuss the development of stress detectors for two distinct types of data: physiological and behavioural. We also address the specific challenges associated with designing effective stress detection systems for each data type and compare the features and classifiers used in both laboratory and real-world contexts. Additionally, this paper proposes future research directions aimed at crafting stress detectors that are robust and effective in real-life scenarios.

Abstract Stress is a common feeling in people’s day-to-day life, especially at work, being the cause of several health problems and absenteeism. Despite the difficulty in identifying it properly, several studies have established a correlation between stress and perceivable human features. The problem of detecting stress has attracted significant attention in the last decade. It has been mainly addressed through the analysis of physiological signals in the execution of specific tasks in controlled environments. Taking advantage of technological advances that allow to collect stress-related data in a non-invasive way, the goal of this work is to provide an alternative approach to detect stress in the workplace without requiring specific controlled conditions. To this end, a video-based plethysmography application that analyses the person’s face and retrieves several physiological signals in a non-invasive way was used. Moreover, in an initial phase, additional information that complements and labels the physiological data was obtained through a brief questionnaire answered by the participants. The data collection pilot took place over a period of two months, having involved 28 volunteers. Several stress detection models were developed; the best trained model achieved an accuracy of 86.8% and a F1 score of 87% on a binary stress/non-stress prediction.

Abstract Imbalanced data are present in various business sectors and must be handled with the proper resampling methods and classification algorithms. To handle imbalanced data, there are numerous resampling and learning method combinations; nonetheless, their effective use necessitates specialised knowledge. In this paper, several approaches, ranging from more accessible to more advanced in the domain of data resampling techniques, will be considered to handle imbalanced data. The application developed delivers recommendations of the most suitable combinations of techniques for a specific dataset by extracting and comparing dataset meta-feature values recorded in a knowledge base. It facilitates effortless classification and automates part of the machine learning pipeline with comparable or better results than state-of-the-art solutions and with a much smaller execution time.

The emergency department of a hospital plays an extremely important role in the healthcare of patients. To maintain a high quality service, clinical professionals need information on how patient flow will evolve in the immediate future. With accurate emergency department forecasts it is possible to better manage available human resources by allocating clinical staff before peak periods, thus preventing service congestion, or releasing clinical staff at less busy times. This paper describes a solution developed for the presentation of hourly, four-hour, eight-hour and daily number of admissions to a hospital’s emergency department. A 10-year history (2009–2018) of the number of emergency admissions in a Portuguese hospital was used. To create the models several methods were tested, including exponential smoothing, SARIMA, autoregressive and recurrent neural network, XGBoost and ensemble learning. The models that generated the most accurate hourly time predictions were the recurrent neural network with one-layer (sMAPE = 23.26%) and with three layers (sMAPE = 23.12%) and XGBoost (sMAPE = 23.70%). In terms of efficiency, the XGBoost method has by far outperformed all others. The success of the recurrent neuronal network and XGBoost machine learning methods applied to the prediction of the number of emergency department admissions has been demonstrated here, with an accuracy that surpasses the models found in the literature.

With the increasing importance of e-commerce across the Internet, the need for software agents to support customers and suppliers in buying and selling goods/services is growing rapidly. It is becoming increasingly evident that in a few years the Internet will host a large number of interacting software agents. Most of them will be economically motivated, and will negotiate a variety of goods and services. It is therefore important to consider the economic incentives and behaviors of economic software agents, and to use all available means to anticipate their collective interactions. This paper addresses this concern by presenting a multiagent market simulator designed for analyzing agent market strategies based on a complete understanding of buyer and seller behaviors, preference models and pricing algorithms, and considering risk preferences. The system includes agents that are capable of improving their performance with their own experience, by adapting to the market conditions. The results of the negotiations between agents are analyzed by data mining algorithms to extract rules that give agents feedback to improve their strategies

We forecast a future in which the global economy and the Internet will host a large number of interacting software agents. Most of them will be economically motivated, and will negotiate a variety of goods and services. It is therefore important to consider the economic incentives and behaviours of economic software agents, and to use all available means to anticipate their collective interactions. This paper addresses this concern by presenting a multi-agent market simulator designed for analysing agent market strategies based on a complete understanding of buyer and seller behaviours, preference models and pricing algorithms. The results of the negotiations between agents will be analysed by data mining tools in order to extract rules that will give the agents feedback to improve their strategies.

With the increasing importance of electronic commerce across the Internet it is becoming increasingly evident that in a few years the Internet will host large numbers of interacting software agents. A vast number of them will be economically motivated, and will exchange a variety of goods and services. It is therefore important to consider the economic incentives and behaviours of economic software agents, and to use every available means to anticipate their collective interactions. We address this concern by presenting a market simulator designed for analysing agent market strategies based on a complete understanding of buyer and seller behaviours, preference model and pricing algorithms.