In the current context of increasingly popular and developed human-computer interaction, machine learning (ML) and data mining are becoming more and more important in various areas, such as image detecting, medicine, and commercial companies. Different researchers from various fields have considered ways to improve the correctness rate for data mining. This paper used data sets from Kaggle and UCI machine learning repository, by applying k-nearest neighbor (KNN), multilayer perceptron (MLP), and decision tree (DT) classifiers to these data sets, the confusion matrix shows the result of the correctness rate for different classifiers under these data sets. As a result, the confusion matrices have shown that data adaptability is based on the data sets characteristics which rely on different classifiers specialties.