Design and Analysis of Automated Machine Learning (AutoML) in PowerBI Application Using PyCaret

Abstract

Massively using conventional machine learning requires many people to understand the process. However, the design of an automatic machine learning system using the Pycaret Library can make it easier for users by reducing some of the operations of machine learning in general. This automated machine learning design is managed in the Power BI application so that the display form can be better understood and easily visualized. By carrying out several method stages in the form of literature studies, data collection, designing classification methods using conventional machine learning and AutoML in Power BI, analysis of results, and reporting/publication. The result of comparison of machine learning and Auto ML: In machine learning, 98 lines of line code are used, while in AutoML, eight lines of code are used. Another advantage derived from using AutoML in PowerBI is the feature to be able to visualize processed datasets so that they can obtain information in the form of diagrams, charts, and others. In testing using the breast cancer dataset, the use of Machine Learning is considered better based on the score obtained. Machine Learning received the highest Accuracy using GaussianNB is 0.94. In testing using the Diabetes dataset, the use of AutoML is considered better based on the score obtained. AutoML received the highest Accuracy using CatBoost Model is 0.777. In testing using the heart failure dataset, the use of AutoML is considered better based on the score obtained. AutoML received the highest Accuracy using Random Forest Model with a score of 0.85.