Demand Forecast: PBOM (Planning BOM)

Demand Forecast: PBOM (Planning BOM)

A Bill of Materials (BOM) is a detailed list that outlines all the raw materials, components, and instructions needed to manufacture a product. It specifies quantities, part numbers, and the assembly sequence, serving as a crucial document for managing production, purchasing, and inventory.

Background: Planning BOM is used for demand forecast and material requirement planning (MRP) when final products are highly configurable.

Problem: Once sale is made, it is imperative to decide which Planning BOM the items sold belong to.  The classification is done by a demand planner.

ML solution:  Use the dataset to train the ML model.  The ML model can predict which Planning BOM the items sold belong to.

  • ML algorithm: multi-class classification
  • ML Platform: AWS, Machine Learning, Multi-class classification.
  • Training Data: last 12 months sale orders, which have 3900 SKUs classified into 26 Planning BOMs. The data is kept in Excel file and uploaded to AWS.
  • Supervised Machine Learning: Once uploaded into AWS, the training dataset is split into two parts: one part for creating ML model and another for evaluating ML model.  One can adjust the train dataset and/or ML parameters; and run the process again.  It is interactive until the evaluation result is acceptable.  Then the ML model is ready for deploring.
  • Prediction with ML Model: 600 SKUs from new sale orders are submitted to ASW ML model to predict which Planning BOM the items sold belong to. The ML model makes 15 errors out of 600 , 2.5% error rate.

You are always welcome to contact us to discuss how to start a project.