My capstone project applies predictive analytics to optimize supply chain operations at Acme Manufacturing. Specifically:

Part 1: The project leverages a range of time series forecasting methods, including CNN-LSTM and Prophet models. This comprehensive approach enables the prediction of aggregate monthly demand for over 17,000 parts across three US and four European manufacturing plants. The forecasts are compared to historical demand and evaluated based on Mean Absolute Error (MAE) and Absolute Value Difference in USD, with the Mixed CNN-LSTM/Prophet model showing superior accuracy, reducing the Absolute Value Difference to 3.15 million USD.

Part 2: Implement linear programming for inventory optimization, balancing holding costs, ordering costs, and shortage costs to achieve optimal stock levels. This method improved inventory availability to 89.3% and decreased on-hand inventory value from $24.36 million to $18.2 million by the end of the forecast period.

Part 3: Employ the K-Nearest Neighbors (KNN) algorithm for ABC classification of inventory items, prioritizing stock based on predicted demand and value. This led to an overall decrease in the number of A and B parts, as well as increased revenue and margin for A and B parts.

This case study’s advantage is the practical application of predictive analytics and machine learning to real company data, demonstrating substantial improvements in demand forecasting accuracy and inventory management efficiency. The project showcases the tangible benefits of data-driven decision-making and advanced analytical methods in optimizing supply chain operations at Acme Manufacturing by comparing the predictive models’ results to actual historical data.