Title: Machine leaning solution based on gradient descent algorithm for improved business process outcomes
Authors: Ivana Dimitrovska; Toni Malinovski; Dane Krstevski
Addresses: Faculty of Informatics, FON University, Skopje, Republic of North Macedonia ' National Bank of the Republic of North Macedonia, Skopje, Republic of North Macedonia ' National Bank of the Republic of North Macedonia, Skopje, Republic of North Macedonia
Abstract: This study aims to provide guidelines that can help organisation identify preconditions before they can employ machine learning, as well as provide evidence that machine learning can be used to improve business process outcomes. It considers supervisory learning as a business learning strategy, and employs machine learning solution based on gradient descent algorithm in large enterprise company in North Macedonia. The solution was designed to streamline the business process, automate the activities, and provide resilience to employees' subjectivity, wrong decisions, and human errors. The machine learning solution was used in production for ten months, including period of changes in the business process, and its average accuracy was 95.018% compared to the employees' decisions. Hence, it verifies the appropriateness of the chosen approach, with predictive accuracy that can be meaningful in practice. Although it is a specific case study, it provides valuable information that organisations can use while undertaking similar initiatives.
Keywords: machine learning; linear regression; gradient descent algorithm; business process improvement.
DOI: 10.1504/IJBIR.2021.114047
International Journal of Business Innovation and Research, 2021 Vol.24 No.4, pp.555 - 570
Received: 09 Nov 2018
Accepted: 06 Jul 2019
Published online: 08 Apr 2021 *