DOI: 10.7763/IJCTE.2026.V18.1400
Towards Mobile-deployable Incident Report Classification: A Statistically Validated Benchmark of Classical Machine Learning Algorithms
Email: aaronpaul.delarosa@bulsu.edu.ph (A.P.M.D.R.); donna.fernando@bulsu.edu.ph (D.Q.F.); rosalyn.reyes@bulsu.edu.ph (R.P.R.); evelyn.samson@bulsu.edu.ph (E.C.S.)
*Corresponding author
Manuscript received October 20, 2025; revised January 30, 2026; accepted May 15, 2026; published September 11, 2026
Abstract—Efficient and accurate classification of incident reports is essential for improving decision-making in local governance systems. In the Philippine context, a barangay, the smallest administrative unit, serves as the frontline in handling community-level incidents. While many studies apply machine learning to text classification, limited attention has been given to computational efficiency and deployment feasibility in resource-constrained environments. This study presents a comparative evaluation of classical machine learning algorithms for classifying real-world barangay incident reports using Term Frequency-Inverse Document Frequency (TF-IDF) features. A lightweight embedding-based baseline (FastText) was also included to assess whether semantic representations improve performance under deployment constraints. The evaluated models span probabilistic, linear, margin-based, online, and ensemble learning paradigms, enabling analysis of predictive performance and computational trade-offs. Linear Support Vector Machine (SVM) achieved the highest performance, with a mean accuracy of 0.6932 ± 0.0408 and a weighted F1-score of 0.6890 ± 0.0425. FastText achieved competitive results (accuracy = 0.6633 ± 0.0388) but did not outperform SVM. Wilcoxon signed-rank testing confirmed that SVM significantly outperformed all models except Stochastic Gradient Descent (SGD) (p < 0.05). Results indicate that linear classifiers provide the best balance between predictive performance and computational efficiency, making them suitable for mobile and resource-constrained deployment.
Keywords—incident report classification, Term Frequency-Inverse Document Frequency (TF-IDF), FastText, machine learning benchmarking, resource-constrained deployment, statistical validation
Cite: Aaron Paul M. Dela Rosa, Donna Q. Fernando, Rosalyn P. Reyes, and Evelyn C. Samson, "Towards Mobile-deployable Incident Report Classification: A Statistically Validated Benchmark of Classical Machine Learning Algorithms," International Journal of Computer Theory and Engineering, vol. 18, no. 3, pp. 193-207, 2026.
Copyright © 2026 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).