Smart Property, Plant, and Equipment Management Using a Random Forest Classification Algorithm

Abstract

Property, plant, and equipment management requires accurate records, reliable depreciation calculations, condition monitoring, and timely lifecycle decisions. This study developed a web-based property, plant, and equipment management system using a Random Forest classification algorithm. The system uses a MySQL database and PHP- and Python-based processing to support asset registration, record retrieval, calculation, classification, and decision-support reporting. It calculates straight-line, double-declining-balance, and sum-of-the-years’-digits depreciation and generates classification-based lifecycle guidance. A dataset of 612 asset records from 2009 to 2025 was used for model development. Inputs included asset category, asset type, purchase date, useful life, maintenance frequency, maintenance cost, and recurring issues. A stratified 80:20 training-test split evaluated two classifiers: asset status (Active, For Maintenance, For Replacement, or For Disposal) and asset outcome (Best, Average, or Worst). Functional testing compared depreciation outputs with expected values, while 11 intended users completed a structured 7-point Likert-scale survey. On the held-out test set, the asset-status model achieved 95.12% accuracy, 95.65% weighted precision, 95.12% weighted recall, 95.16% weighted F1-score, and 99.47% multiclass ROC-AUC. The asset-outcome model achieved 96.75% accuracy, 96.94% weighted precision, 96.75% weighted recall, 96.67% weighted F1-score, and 99.76% multiclass ROC-AUC. The depreciation outputs matched expected values, and user-acceptance construct means ranged from 6.30 to 6.80 out of 7. The findings demonstrate the feasibility and initial effectiveness, within the evaluated setting, of integrating asset records, depreciation, predictive classification, and reporting to support asset lifecycle decisions. The outputs are decision-support information and do not replace professional judgment or organizational approval. This supports more consistent review of asset condition and value.