Improve Extremely Fast Decision Tree Performance through Training Dataset Size for Early Prediction of Heart Diseases

Training dataset characteristics can have an important impact on the performance of the classification models. One of the major characteristics of the training data that affects the accuracy of a classification model is the size. A lot of studies demonstrate that classification algorithms require large datasets to train effective models. However, large datasets, especially with high-quality labels, can be expensive to obtain. This paper proposes practical techniques to improve the accuracy measure with a small training dataset by using two online classification models: Very Fast Decision Tree and Extremely Fast Decision Tree. The use case of our study is the early prediction of cardiac diseases. The experiments showed that with a parameter estimation method in the models and balancing of classes of the training dataset, we can get similar performances of large data in a smaller dataset. We observe that dataset size has a conspicuous lack of effect on the training of these models, undoubtedly there are other ways of extrapolating online decision tree algorithm's performance with a small dataset.