Comparison of Machine Learning Services

Abstract

In this paper, three Cloud Machine Learning services are considered using the same dataset to run predictions on each of them. These services are quantitatively and qualitatively analyzed and compared by considering their mode of operation, data processing, prediction creation, model creation, cost, and algorithm. Although Google Prediction API offers fast model training and model creation as compared to Windows Azure Machine Learning Studio and Amazon Machine Learning; it has lesser visualization tools. The focus of this thesis is to provide better understanding on how these services operate, identify important research directions in Machine Learning field, and present a clear picture of the functionalities of the services to aid the decision of developers when choosing which service will best suit their Machine Learning solutions.