Journal of Transactions in Systems Engineering

Unsupervised Clustering of Multivariate Sports Activity Data Using K-Means: A Study on the Sport Data Multivariate Time Series Dataset

Ahmed T. Alhasani — 2025-07-04

This work investigates the combination of unsupervised machine learning with blockchain- influenced data integrity aspects on multivariate time series (MTS) sports activity data. Using the SportData MTS dataset with complex physiological and movement parameters such as heart rate, speed, and altitude, we used K-Means clustering to uncover hidden patterns in the data and incorporated blockchain-influenced hash chains for traceability and integrity of data. Each of the datasets was standardized to ensure equal scaling, and three clusters were identified using silhouette score and elbow method evaluation. The result confirms K-Means to effectively cluster the data into tightly separated groups, with principal component analysis (PCA) plots confirming that there is substantial separation. Silhouette score analysis also confirmed the compactness and separability of groups. In addition, blockchain-inspired hashing was applied to each record to simulate tamper- evidence, providing a firm grounding for secure machine learning pipelines. The end-to-end solution not only reveals the inherent structure in sports activity data but also hints at maintaining data integrity to provide sound and transparent machine learning results, paving the way for future work in secure sports analytics, activity recognition, and anomaly detection.

A Fusion-Based Machine Learning Framework for Lung Cancer Survival Prediction Using Clinical and Lifestyle Data

Hussein Alkattan — 2025-07-08

Lung cancer is one of the deadliest diseases worldwide, highlighting the criticality of precise survival prediction models. This work proposes an exhaustive fusion-based machine learning approach for lung cancer survival prediction using heterogeneous features such as clinical indicators, demographic information, and lifestyle factors. A publicly available dataset of more than 800,000 records was pre-processed, statistically analysed, and dimensionally reduced for computational tractability. Feature-level fusion was used to merge multivariate features, after which decision-level fusion was implemented through soft voting ensembles. Five fusion configurations using Logistic Regression, Random Forest, Support Vector Machine, k-Nearest Neighbours, and Naive Bayes classifiers were evaluated. It was noted that the simpler combinations like Logistic Regression and Random Forest worked better than larger ensembles, with accuracy of 70% and AUC of 0.61 after class balancing. Correlation and statistical analysis also showed weak linear relationships with survival, underscoring the need for non-linear modelling strategies. Every fusion model was assessed with ROC curves and confusion matrices, providing an overall view of prediction strength. The study demonstrates that fusion techniques can significantly improve survival prediction in lung cancer patients and can be the foundation for actual clinical decision support systems.

Hybrid Machine‐Learning Framework for Predicting Student Placement

Ahmed Hamid Elias — 2025-07-18

Accurate prediction of the results of college student placement can help institutions detect potential risk students and tailor career-readiness interventions. In this study, using a publicly available dataset of 10,000 students with eight predictors intelligence quotient (IQ), previous semester's performance, cumulative grade point average (CGPA), academic rating, internship or not, extra-curricular score, communication skills, and projects completed, we develop and validate a hybrid stacking ensemble classifier. After numerical feature standardization and binary category encoding, we trained three base learners (support vector machine, random forest, and logistic regression) and combined them with a logistic regression meta-learner. Comparative experiments on an 80/20 train–test split show that the stacking ensemble outperforms individual models, with 100 % accuracy, precision, recall, and F1-score on the test set, whereas logistic regression alone attained 90.4 % accuracy. A correlation analysis declares CGPA and performance in the previous semester as the single best predictors for placement. Receiver operating characteristic (ROC) curves and confusion matrices also confirm the greater discrimination power and stability of the ensemble. All these results confirm that stacking heterogeneous classifiers provides a stable and interpretable approach to student placement prediction, with potential use in academic advising and early warning systems.