Multiclass Classification with Decision Trees, Naive Bayes and Logistic Regression: An Application with R

Bilgic E., Esen M. F.

International Conference on Data Science, Machine Learning and Statistics, Van, Türkiye, 26 - 29 Haziran 2019, ss.280-282

  • Basıldığı Şehir: Van
  • Basıldığı Ülke: Türkiye
  • Sayfa Sayıları: ss.280-282


The developments in Data Science have also enabled the emergence of data which can support the decision making processes of the companies. Data Mining (DM), which includes the processes of data acquisition, storage and analysis, has been successfully implemented in different business problems. In this study, the size of the firms that the shares belong to will be estimated by classification analysis by means of variables such as the type of the transaction, quantity and amount of the shares. This analysis is thought to be useful in exploring investor behaviors. In this context, three different classification techniques, Decision Trees (DT), Logistic Regression (LR) and Naïve Bayes (NB) were applied to a dataset of financial transactions using R programming language. As a result of the study, when the performance of the classification techniques was compared, it was found that the DT technique made 72 % correct classification.