Predicting the rainfall status of a region has a great impact on certain factors, such as arranging agricultural activities, enabling efficient water planning, and taking precautionary measures for possible disasters (flood/drought). Due to the seriousness of the subject, the timely and accurate prediction of rainfall is highly desirable and critical for environmentally sustainable development. In this study, an ensemble of K-stars (EK-stars) approach was proposed to predict the next-day rainfall status using meteorological data, such as the temperature, humidity, pressure, and sunshine, that were collected between the years 2007 and 2017 in Australia. This study also introduced the probability-based aggregating (pagging) approach when building and combining multiple classifiers for rainfall prediction. In the implementation of the EK-stars, different experimental setups were carried out, including the change of input parameter of the algorithm, the use of different methods in the pagging step, and whether the feature selection was performed or not. The EK-stars outperformed the original K-star algorithm and the recently proposed studies in terms of the classification accuracy by making predictions that were the closest to reality. This study shows that the proposed method is promising for generating accurate predictions for the sustainable development of environmental systems.

Traditional indoor human activity recognition (HAR) has been defined as a time-series data classification problem and requires feature extraction. The current indoor HAR systems still lack transparent, interpretable, and explainable approaches that can generate human-understandable information. This paper proposes a new approach, called Human Activity Recognition on Signal Images (HARSI), which defines the HAR problem as an image classification problem to improve both explainability and recognition accuracy. The proposed HARSI method collects sensor data from the Internet of Things (IoT) environment and transforms the raw signal data into some visual understandable images to take advantage of the strengths of convolutional neural networks (CNNs) in handling image data. This study focuses on the recognition of symmetric human activities, including walking, jogging, moving downstairs, moving upstairs, standing, and sitting. The experimental results carried out on a real-world dataset showed that a significant improvement (13.72%) was achieved by the proposed HARSI model compared to the traditional machine learning models. The results also showed that our method (98%) outperformed the state-of-the-art methods (90.94%) in terms of classification accuracy.

As one of the entropy-based methods, the k-Star algorithm benefits from information theory in computing the distances between data instances during the classification task. k-Star is a machine learning method with a high classification performance and strong generalization ability. Nevertheless, as a standard supervised learning method, it performs learning only from labeled data. This paper proposes an improved method, called Semi-Supervised k-Star (SSS), which makes efficient predictions by considering unlabeled data in addition to labeled data. Moreover, it introduces a novel semi-supervised learning approach, called holo-training, against self-training. It has the advantage of enabling a powerful and robust model of data by combining multiple classifiers and using an entropy measure. The results of extensive experimental studies showed that the proposed holo-training approach outperformed the self-training approach on 13 out of the 18 datasets. Furthermore, the proposed SSS method achieved higher accuracy (95.25%) than the state-of-the-art semi-supervised methods (90.01%) on average. The significance of the experimental results was validated by using both the Binomial Sign test and the Friedman test.

The aim of this study is to develop a new approach to be able to correctly predict the outcome of electronic sports (eSports) matches using machine learning methods. Previous research has emphasized player-centric prediction and has used standard (single-instance) classification techniques. However, a team-centric classification is required since team cooperation is essential in completing game missions and achieving final success. To bridge this gap, in this study, we propose a new approach, called Multi-Objective Multi-Instance Learning (MOMIL). It is the first study that applies the multi-instance learning technique to make win predictions in eSports. The proposed approach jointly considers the objectives of the players in a team to capture relationships between players during the classification. In this study, entropy was used as a measure to determine the impurity (uncertainty) of the training dataset when building decision trees for classification. The experiments that were carried out on a publicly available eSports dataset show that the proposed multi-objective multi-instance classification approach outperforms the standard classification approach in terms of accuracy. Unlike the previous studies, we built the models on season-based data. Our approach is up to 95% accurate for win prediction in eSports. Our method achieved higher performance than the state-of-the-art methods tested on the same dataset.