An autoML algorithm to select suitable decision tree algorithm for classification

This paper presents an automatic machine learning (autoML) algorithm to select a decision tree algorithm which is most suitable for the stated requirements by the user for classification. The user can specify requirements in terms of algorithmic characteristics such as speed, space, scalability and interpretability of the decision tree algorithms for selection. The user can also specify any missing values in the dataset and data types (numeric or categorical or mixed) in the dataset that will be used for training and testing of the selected decision tree algorithm. Based on these inputs the autoML algorithm selects one of the few best decision tree algorithms which are included for automatic selection. The decision tree algorithms available for selection include C5.0, C4.5, ID3, CART and CHAID. The autoML algorithm is tested for the selection of the suitable decision tree algorithm based on algorithmic characteristics, and data types and missing values in the dataset as fed by the user.