Research on Business Design Decisions Driven by Machine Learning

Abstract

In the current fast-paced, competitive market, artificial intelligence (AI)-driven decision-making has become an indispensable part, sparking intense interest in industrial machine learning (ML) applications. The demand for experts in requirements analysis far outstrips supply, and one solution is to enhance the user-friendliness of machine learning frameworks. Automated Machine Learning (AutoML) is seen as an attempt to address the issue of expertise shortage by providing fully automated, customized solutions. This study aims to analyze the potential of AutoML in business analytics applications, with the goal of promoting the widespread application of ML. The H2O AutoML framework demonstrated excellent performance, robustness, and reliability by benchmarking against manually tuned stacked ML models on three real datasets. This framework is fast, easy to use, and provides reliable results close to those of professionally tuned ML models. Currently, the capabilities of the H2O AutoML framework can support rapid prototyping, shortening the development and deployment cycle, bridging the gap between the supply and demand of ML experts, and marking a significant step towards fully automated decision-making in business analytics..