LEVERAGING MACHINE LEARNING MODELS TO BOOST THE MANIPULATION OF UNSTRUCTURED DECISIONS AND UNSTRUCTURED DATA IN DIGITAL BUSINESS FIRMS

Abstract

In today's fast-paced digital landscape, businesses are navigating a world filled with unstructured decisions where information is often incomplete, uncertainty looms large, and change is a constant companion. The main objective of this research is to explore and investigate how machine learning (ML) models can step in to extract valuable insights from vast and diverse datasets, ultimately enhancing decision-making processes for the unstructured decisions to minimize the failures of business firm projects. The research methodology employs a mixed-methods strategy, blending a review of recent literature with two in-depth case studies analysis; JPMorgan Chase in finance and Airbus in aviation. The results reveal that ML made a significant difference in both scenarios. At JPMorgan, the accuracy of fraud detection soared from 70 % to 92 %, while false positives plummeted from 50 % to a mere 8 %. What once took up to two days to uncover now happens in under a minute, leading to a reduction in operational costs by around 35 %. Similarly, Airbus experienced notable gains in a different domain: ML-driven predictive maintenance slashed unexpected breakdowns by 60 %, cut maintenance expenses by 40%, and boosted fleet availability by 29 %. This analysis clearly illustrates and concludes that ML is here to support human experts to solve the unstructured decisions, not replace them. The research also highlights future challenges and priorities, such as the necessity for explainable AI (XAI), enhanced data governance, and improved integration of ML with human decision-making to ensure that its adoption remains both effective and responsible.