Simplifying Decision Tree Classification Through the AutoDTrees Web Application and Service

Various algorithms are utilized for the purpose of classification, with Decision Trees being one of the most popular. This is due to their easily understandable structure and simple way of operation, which led to their adoption in a variety of applications. However, the usage of Decision Trees often becomes challenging due to the fact that users need to be familiar with Machine Learning, have programming skills or knowledge of specialized scientific software. The present paper aims to address these issues by presenting AutoDTrees, a web-based application that offers the ability to utilize the Decision Trees in a simple and fast way. This can be done through a user-friendly interface, while an open-source Web API is provided for developers. AutoDTrees allows any user to select the preferred dataset and define parameters, in order to build Decision Tree models. Then, the model effectiveness can be evaluated by using the k-fold cross-validation and presenting detailed metrics. Users are then able to save the pre-trained model and reuse it for predicting unclassified instances or visualizing the Decision Tree. AutoDTrees was evaluated in terms of user experience using the System Usability Scale (SUS), with the results indicating that it can be a useful tool for a wide range of users, regardless of their experience level.