Machine Learning in Predictive Quality Management

Abstract

This thesis investigates how machine learning techniques has greater advantages over traditional Statistical Process Control (SPC) methods and Anomaly Detection (AD) in monitoring and controlling process parameters so that the process remains stable and in control to meet required quality satisfactions. The broader term of Predictive Quality Management introduces an important subject of predictive analytics which can be collaborated with machine learning tools to predict or forecast future process patterns. This thesis employs a case study, of fishing industries that amass large marine data in form of Automatic Identification Systems data, catch data and environmental but fail to draw meaningful correlations between data variables towards sustainable and efficient use of fishing vessel resources. Different machine learning models were implemented, techniques for data cleaning and preprocessing provided a leeway to draw patterns and trends in our dataset. and a performance evaluation using suitable metrics was conducted to determine which regressor algorithm predicts and generate forecasts for location of fish species. This thesis contributes to a deep understanding of data analysis and offers recommendations to decision makers.