Öznitelik Seçim Yöntemlerinin Toplam Ekipman Etkinliği Tahmin Başarısı Üzerindeki Etkisinin Araştırılması

Overall equipment effectiveness (OEE) describes production efficiency by combining availability, performance, and quality is used to evaluate equipment’s performance. This research’s aim investigate the potential of feature selection techniques multiple linear regression method, which one machine learning techniques, in successfully predicting OEE corrugated department a box factory. In study, six different planned downtimes information on seventeen previously known concepts related activities be performed are as input features. Moreover, backward elimination, forward selection, stepwise correlation-based (CFS), genetic algorithm, random forest, extra trees, ridge regression, lasso elastic net methods proposed find most distinctive subset dataset. As result analyses data set consisting 23 features, 1 output 1204 working days information, - model, selects 19 attributes, gave best average R2 value compared other models developed. Occam's razor principle taken into account since there not great difference between values obtained. Among developed according principle, model yielded among those that selected fewest