Backorder Prediction

Published: 3 September 2019| Version 1 | DOI: 10.17632/krnbcxksn3.1
Contributor:
Rodrigo De Santis

Description

The dataset contains historical data for inventory-active products from the previous 8 weeks of the week we would like to predict, captured as a photo of all inventory at the beginning of the week. Attributes SKU: Unique material identifier; INV: Current inventory level of material; TIM: Registered transit time; FOR-: Forecast sales for the next 3, 6, and 9 months; SAL-: Sales quantity for the prior 1, 3, 5, and 9 months; MIN: Minimum recommended amount in stock (MIN); OVRP: Parts overdue from source; SUP-: Supplier performance in last 1 and 2 semesters; OVRA: Amount of stock orders overdue (OVRA); RSK-: General risk flags associated to the material; BO: Product went on backorder. Evaluation Metrics We applied Area Under Receiver Operator Curve (AUROC) for primary evaluation, and Precision-Recall curves for post-analysis.

Files

Categories

Machine Learning, Inventory Management, System Fault Detection

License