Text Mining Algorithm for Equipment Inspection Reports via AppSheet Inspection System

This research aims to develop and analyze a risk classification model for assets using data from the 2023 fiscal year asset inspection report via the AppSheet platform, consisting of 7,000 records. The analysis, performed in RapidMiner, utilized variables such as asset type, lifespan, value, and maintenance records. K-Means Clustering grouped risks into three levels: low, medium, and high, followed by Risk Index calculation. Model performance was evaluated using 10-Fold Cross Validation, comparing six algorithms: Deep Learning, Gradient Boosted Trees, Naive Bayes, Random Forest, SVM, and XGBoost. XGBoost outperformed others with 99.39% accuracy, 0.99 Kappa, 0.61% error rate, 99.32% weighted recall, and 99.40% weighted precision. Results were visualized through a Google Looker Studio dashboard, supporting efficient risk classification and decision-making.