Utilizing machine learning to enhance optimal inventory management : case : aviation industry spart parts

Abstract

This thesis addresses inventory management challenges faced by global aerospace platforms. The special focus is given to a case company, GA Telesis which is a full-service aircraft maintenance and component services provider. Strategies for optimizing inventory levels are explored, including classifying items based on historical usage, cost and assembled engine models to identify critical items and utilizing ABC/XYZ analyses alongside machine learning for inventory management. The quantitative study methodology employs an action research approach with machine learning models of Autoregressive Integrated Moving Average (ARIMA), Support Vector Machines (SVM), Linear Regression (LR), Gradient Boosting (GB) and statistical techniques, including Mean Error (ME), Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE) to predict future demand based on numerical data sourced from the IFS system's procurement records of the case company. Additionally, interviews with key personnel provide insights into real-world challenges faced by the company, allowing the study to propose solutions specifically tailored to address these issues. The key finding of this study is that the ARIMA model demonstrated superior performance in demand forecasting, corroborating with existing literature and validating its effectiveness in the aviation sector.