Computer Fault Classification Using Support Vector Machine in Orange Data Mining

Abstract

Computer repair services at small and medium enterprises (SMEs) often struggle to estimate damage categories before physically inspecting devices, reducing service efficiency and customer transparency. This study develops a Support Vector Machine (SVM) classification model implemented through Orange Data Mining to predict device damage categories from initial customer-reported information. The dataset comprises 58 service transactions from a Bandung-based repair business. Preprocessing included keyword matching to convert free-text complaints into structured categories, yielding five input features. Evaluated via 10-fold stratified cross-validation, the model achieved accuracy, AUC, and F1-Score above 0.89, with physical condition and complaint type as the most influential features. Results demonstrate that SVM via Orange Data Mining is effectively applicable to SME-scale operational data, enabling earlier damage estimation and improved service workflow.