SPECIFIC MACHINE LEARNING ALGORITHMS AS EFFICIENT SOLUTIONS FOR COMPLEX BUSINESS PROCESSES

This paper presents and critically investigates some unsupervised and supervised algorithms specific to machine learning, which are particularly suited for business processes and their high dynamical evolution. We particularly highlight the general merits of machine learning and mathematical foundation in the big data era and its importance for applications to industrial and business processes, in the context of well- defined economic problems, with their variables and, especially, the predictive functions to be solved. Then, the paper discusses some existing algorithms and their mathematical formulations and merits, based on existing literature and recent usage in various economic applications. Such machine learning algorithms form today a calculation distribution framework, available in open-source software for storing data and running applications such as Handoop. We also reveal the importance of robustness and of data quality, as many of the business ecosystems are currently filled with dirty and useless data, Finally, we suggest that for precisely observing, quantifying and predicting the performance progress of organizational management, transparency for the users of machine learning is required, as opposed to traditional rather opaque machine learning processes. Finally, the proposed algorithms overview and discussion is helpful to provide first steps in learning how to apply machine learning to make your business more efficient, more effective and more profitable.