A Survey of The State-Of-The-Art AutoML Tools and Their Comparisons

Abstract

Machine learning is used effectively in many areas today and its usage area is increasing day by day. In addition, processes based on machine learning are also developing in a technology-oriented manner, and users are gaining new perspectives on solving current problems. While machine learning makes predictions about stocks in the financial sector, it also plays an active role in early diagnosis of diseases in the healthcare sector. It is actively used in route calculation and defective product detection in the field of production and logistics, and in situations such as analysis of customer behavior and product recommendations in the shopping sector. AutoML can be defined as a process that aims to automate the machine learning process end-to-end. It enables the machine learning process to be accelerated by automating especially time-consuming tasks that work with the logic of repetition, and it allows people who work in this field to create more efficient and productive models. In addition, AutoML helps users who are not experts in this field in the stages of machine learning model development, data management, analysis and evaluation of their own data, by providing various conveniences to users in model training and subsequent stages. In this article, after discussing what AutoML is, AutoML processes and areas of use, information about various AutoML platforms have been given, the differences between widely used AutoML platforms will be evaluated, and their advantages and disadvantages compared to each other will be included.