Automated Machine Learning Model Selection Analysis

Machine learning (ML) is a branch of artificial intelligence (AI) in which a program uses large sets of data to “learn” methods and patterns. Following the learning process, a model is created based on the learned information, which can then be applied to new datasets to automatically perform tasks. Automated Machine Learning (AutoML) is a new tool which can automate the mundane tasks involved in machine learning, such as data preprocessing, model selection, and hyper parameter tuning. Model selection is one of the most important aspects in creating an effective ML model as it affects the algorithms used to perform ML. AutoML will automate the process of trial and error needed to find the most effective model for a task, however, there are different methods used to select a model. This paper will investigate the various tools used to perform AutoML, taking note of the methods used to for model selection to find the current state of model selection in AutoML.