Root cause analysis of familiarity biases in classification of inventory items based on logical patterns recognition

• Inventory classification done by experts is biased by familiarity and experience.
• Familiarity bias will lead to non-accurate further classifications.
• Familiarity bias is detected by multi-class model of Logical Analysis of Data.
• Bias is corrected by pattern recognition analysis using LAD multi-class model.
• Pattern-based classification is effective for complex processes’ characterization.

Multi-features ABC inventory classification (MCIC) is targeted to optimize inventory management through inventory items classification in order to set policies and rules to manage them. A flawed classification of items may lead to financial losses and customers dissatisfaction. This paper presents a technique for the identification and the correction of the bias that exists in the ABC items classification done by inventory experts and decision makers. In this study, the classification familiarity bias is found and rectified through patterns recognition. A pattern based reclassification is proposed using a multi-class model based on Logical Analysis of Data (LAD). Accuracy prediction tests are conducted in order to evaluate the proposed pattern based classification. The results are compared with the classification obtained based on the Euclidean distance.