Leveraging Information Technology and Machine Learning for Explainable Business Intelligence in Quality Prediction

In the current data-driven corporate environment, companies understand that the main source of gaining a competitive edge is not through data ownership but through the capability to analyse data fast and correctly to generate actionable insights. This article thoroughly outlines the synergy between information technology infrastructure and machine learning algorithms in the production of valuable business intelligence. Based on an apple quality dataset, which contains 4, 000 instances with 8 physicochemical attributes, we built and evaluated three supervised machine learning classifiers, Logistic Regression, K-Nearest Neighbour and Nave Bayes. All three models are evaluated using four regression metrics: precision recall F1-score and accuracy. Experimental results show that K-Nearest Neighbour is not only the most accurate (90%) classifier among all three, but the best-performing classifier for this set of data. Logistic Regression and Nave Bayes have equal accuracy of 75%. These results show that, KNN is the best classifier suitable for prediction of quality research says under consideration. On top of machine learning model comparison, this paper also discusses several information technology issues encountered when applying machine learning algorithms like data flow architecture, feature engineering, model selection, as well as how it can be integrated with an existing business intelligence system for use. It is reasonable to assume that businesses may use this knowledge to take predictive actions to avoid poor quality. This study has added a new insight to the expanding domain at the crossroad of IT infrastructure, applications of machine learning algorithms, explainable artificial intelligence, and business analytics by providing standards of theory and practice that can be used to develop earlier tests and applications.