Custom Plastic Molding Job Shop
Client Challenge
A CPQ solution was needed for growing demand and complexity of custom products
The client believed no commonality existed across historical jobs — every quote built from scratch
Client Solution
Hypothesis Testing: Established the alternative hypothesis (≤10 production profiles) against the client’s null
Data Mining: Converted S2K production history to SQL Server; used SSIS Term Extraction and Term Lookup to re-tag routing descriptions and the Data Profiling Task to expose distributions and outliers, then applied SSAS clustering, neural networks, Naive Bayes, decision trees, and basket analysis across millions of historical records (January 2011 – February 2012, three plant locations)
Design: Future-state Quote-to-Cash and PLM processes; PDM best-practice documentation; recommended Master Planner and PDM Manager roles
Client Impact Results
Result: Null hypothesis rejected — 5 core profiles generated by ML now anchor the proposed CPQ application
Adoption: Future-state QTC and PLM approved by C-level and departmental leadership
Client Technologies
SQL Server 2008 Integration Services (Term Extraction, Term Lookup, Data Profiling)
SQL Server Analysis Services data mining (clustering, neural networks, Naive Bayes, decision trees, basket analysis) with the Excel Data Mining add-in
S2K ERP
Data mining methodology
