An Ordinal Logistic Regression Model for Class Prediction of Inventory Items

Effective inventory management needs proper inventory classification, which optimizes stock control and resource allocation. This study develops an Ordinal Logistic Regression (OLR) model based on three crucial factors—Annual Dollar Usage (ADU), Average Unit Cost (AUC), and Lead Time (LT)— to classify inventory items based on priority levels of classes. A sample dataset was generated by defining ranges for the independent variables and converted into ordinal data as the input of the formulated ORL model, whereas the dependent variable (Class) was assigned based on predefined criteria. The analysis revealed a significant relationship between the independent and dependent variables, indicating that higher independent variable values lead to higher-priority classifications. The developed OLR model demonstrated a high accuracy of 82.98% in inventory classification of the test dataset, correctly categorizing 39 out of 47 items. Furthermore, model performance evaluation yielded overall accuracy, macro precision, and weighted average precision values of 0.8865, 0.8197, and 0.8537, respectively—all exceeding 80%. These results indicate that the proposed OLR model is highly effective and can be practically applied to inventory classification.