Georgia Tech SCL Georgia Tech SCL Georgia Tech SCL

• How ML relates to SCM
• ML algorithms such as regression trees, clustering techniques, decision trees, random forests, logistic regression
• Aspects of ML projects including parameter tuning, cross validation, and assess model performance
• Application of ML in demand forecasting for sales and operations planning (S&OP) and inventory management
• Application of ML in predictive maintenance
• Hands-on practice with these skills using data from the (fictional) Cardboard Company (CBC)

Below is format for the online/virtual-Instructor led version of the course.

Webinar 1 – Machine Learning for SCM (regression and clustering)

Intro to ML as it relates to SCM

Regression and clustering models

• Regression trees
• Advanced time series forecasting
• Various clustering techniques

Assessing model performance

• Interpretability vs accuracy
• Parameter tuning
• Cross validation

Activity: Use techniques to forecast the Cardboard Company (CBC)’s demand

Homework: Finish CBC’s demand forecast
 

Webinar 2 – Demand Forecasting: S&OP and Inventory Management

Perform customer segmentation

Use ML to determine inventory policies

Activity: Determine CBC’s inventory polices using ML

Homework: Finish creating and assessing CBC’s ML models
 

Webinar 3 – Machine Learning for SCM (classification)

Classification models

• Decision trees
• Random forests
• Logistic regression

Building planning tools with large and real-time data sets

Activity: Apply these models as they relate to CBC’s SCM

Homework: Finish creating CBC classification models
 

Webinar 4 – Production Planning and Predictive Maintenance

Production planning

• Building anomaly detection models
• Activity: build an anomaly detection model to sensor CBC’s paper production data

Predictive maintenance

• Using models to support proactive production maintenance planning
• Activity: build a predictive maintenance model for CBC

Final assignment (due 1 week after last webinar)

• Complete the CBC predictive maintenance model
• Create an explanation of your model choices for the instructors
• Complete final assessment