Information and Communication Technologies fueled social networking and facilitated communication. However, cyberbullying on the platform had detrimental ramifications. The user-dependent mechanisms like reporting, blocking, and removing bullying posts online is manual and ineffective. Bag-of-words text representation without metadata limited cyberbullying post text classification. This research developed an automatic system for cyberbullying detection with two approaches: Conventional Machine Learning and Transfer Learning. This research adopted AMiCA data encompassing significant amount of cyberbullying context and structured annotation process. Textual, sentiment and emotional, static and contextual word embeddings, psycholinguistics, term lists, and toxicity features were used in the conventional Machine Learning approach. This study was the first to use toxicity features to detect cyberbullying. This study is also the first to use the latest psycholinguistics features from the Linguistic Inquiry and Word (LIWC) 2022 tool, as well as Empath’s lexicon, to detect cyberbullying. The contextual embeddings of ggeluBert, tnBert, and DistilBert have alike performance, however DistilBert embeddings were elected for higher F-measure. Textual features, DistilBert embeddings, and toxicity features that struck new benchmark were the top three unique features when fed individually. The model’s performance was boosted to F-measure of 64.8% after feeding with a combination of textual, sentiment, DistilBert embeddings, psycholinguistics, and toxicity features to the Logistic Regression model that outperforms Linear SVC with faster training time and efficient handling of high-dimensionality features. Transfer Learning approach was by fine-tuning optimized version Pre-trained Language Models namely, DistilBert, DistilRoBerta, and Electra-small which were found to have speedier training computation than their base form. The fine-tuned DistilBert resulted with the highest F-measure of 72.42%, surpassing CML. Our research concluded that TL was the best for uplifted performance and lesser effort as feature engineering and resampling was omitted.

Background: Cardiovascular disease is the leading cause of death in many countries. Physicians often diagnose cardiovascular disease based on current clinical tests and previous experience of diagnosing patients with similar symptoms. Patients who suffer from heart disease require quick diagnosis, early treatment and constant observations. To address their needs, many data mining approaches have been used in the past in diagnosing and predicting heart diseases. Previous research was also focused on identifying the significant contributing features to heart disease prediction, however, less importance was given to identifying the strength of these features.Method: This paper is motivated by the gap in the literature, thus proposes an algorithm that measures the strength of the significant features that contribute to heart disease prediction. The study is aimed at predicting heart disease based on the scores of significant features using Weighted Associative Rule Mining.Results: A set of important feature scores and rules were identified in diagnosing heart disease and cardiologists were consulted to confirm the validity of these rules. The experiments performed on the UCI open dataset, widely used for heart disease research yielded the highest confidence score of 98% in predicting heart disease.Conclusion: This study managed to provide a significant contribution in computing the strength scores with significant predictors in heart disease prediction. From the evaluation results, we obtained important rules and achieved highest confidence score by utilizing the computed strength scores of significant predictors on Weighted Associative Rule Mining in predicting heart disease.

The selection of a meta-learner determines the success of a stacked ensemble as the meta-learner is responsible for the final predictions of the stacked ensemble. Unfortunately, in imbalanced classification, selecting an appropriate and well-performing meta-learner of stacked ensemble is not straightforward as different meta-learners are advocated by different researchers. To investigate and identify a well-performing type of meta-learner in stacked ensemble for imbalanced classification, an experiment consisting of 19 meta-learners was conducted, detailed in this paper. Among the 19 meta-learners of stacked ensembles, a new weighted combination-based meta-learner that maximizes the H-measure during the training of stacked ensemble was first introduced and implemented in the empirical evaluation of this paper. The classification performances of stacked ensembles with 19 different meta-learners were recorded using both the area under the receiver operating characteristic curve (AUC) and H-measure (a metric that overcomes the deficiencies of the AUC). The weighted combination-based meta-learners of stacked ensembles have better classification performances on imbalanced datasets when compared to bagging-based, boosting-based, Decision Trees, Support Vector Machines, Naive Bayes, and Feedforward Neural Network meta-learners. Thus, the adoption of weighted combination-based meta-learners in stacked ensembles is recommended for their better performance on imbalanced datasets. Also, based on the empirical results, we identified better-performing meta-learners (such as the AUC maximizing meta-learner and the H-measure maximizing meta-learner) than the widely adopted meta-learner - Logistic Regression - in imbalanced classification.

The Myers-Briggs Type Indicator (MBTI) is a well-known personality test that assigns a personality type to a user by using four traits dichotomies. For many years, people have used MBTI as an instrument to develop self-awareness and to guide their personal decisions. Previous researches have good successes in predicting Extraversion-Introversion (E/I), Sensing-Intuition (S/N) and Thinking-Feeling (T/F) dichotomies from textual data but struggled to do so with Judging-Perceiving (J/P) dichotomy. J/P dichotomy in MBTI is a non-separable part of MBTI that have significant inference on human behavior, perception and decision towards their surroundings. It is an assessment on how someone interacts with the world when making decision. This research was set out to evaluate the performance of the individual features and classifiers for J/P dichotomy in personality computing. At the end, data leakage was found in dataset originating from the Personality Forum Café, which was used in recent researches. The results obtained from the previous research on this dataset were suggested to be overly optimistic. Using the same settings, this research managed to outperform previous researches. Five machine learning algorithms were compared, and LightGBM model was recommended for the task of predicting J/P dichotomy in MBTI personality computing.

Background: Dengue fever is a widespread viral disease and one of the world’s major pandemic vector-borne infections, causing serious hazard to humanity. The World Health Organisation (WHO) reported that the incidence of dengue fever has increased dramatically across the world in recent decades. WHO currently estimates an annual incidence of 50–100 million dengue infections worldwide. To date, no tested vaccine or treatment is available to stop or prevent dengue fever. Thus, the importance of predicting dengue outbreaks is significant. The current issue that should be addressed in dengue outbreak prediction is accuracy. A limited number of studies have conducted an in-depth analysis of climate factors in dengue outbreak prediction. Methods: The most important climatic factors that contribute to dengue outbreaks were identified in the current work. Correlation analyses were performed in order to determine these factors and these factors were used as input parameters for machine learning models. Top five machine learning classification models (Bayes network (BN) models, support vector machine (SVM), RBF tree, decision table and naive Bayes) were chosen based on past research. The models were then tested and evaluated on the basis of 4-year data (January 2010 to December 2013) collected in Malaysia. Results: This research has two major contributions. A new risk factor, called the TempeRain factor (TRF), was identified and used as an input parameter for the model of dengue outbreak prediction. Moreover, TRF was applied to demonstrate its strong impact on dengue outbreaks. Experimental results showed that the Bayes Network model with the new meteorological risk factor identified in this study increased accuracy to 92.35% for predicting dengue outbreaks. Conclusions: This research explored the factors used in dengue outbreak prediction systems. The major contribution of this study is identifying new significant factors that contribute to dengue outbreak prediction. From the evaluation result, we obtained a significant improvement in the accuracy of a machine learning model for dengue outbreak prediction.