In this paper, we present an overview of the imbalanced data classification and the difficulties encountered in current approaches, from which we propose a new method, SMOTE-PLS. To evaluate the effectiveness of this new method, we conducted experiments based on standard cancer data sets from UCI sources, including breast-p, coil2000, leukemia, colon-cancer, and yeast. | HNUE JOURNAL OF SCIENCE Natural Sciences 2020 Volume 65 Issue 4A pp. 42-50 This paper is available online at http A NEW HYBRID METHOD TO IMPROVE THE EFFECTIVENESS OF CANCER DATA CLASSIFICATION Nguyen Thi Chinh 1 Dao Thi Minh2 Le Xuan Ly3 and Dang Xuan Tho 4 1 Gifted High School Hanoi National University of Education 2 Natural Science department Thai Binh Teacher Training College 3 School of Applied Mathematics and Informatics Hanoi University of Science and Technology 4 Faculty of Information Technology Hanoi National University of Education Abstract. Imbalanced data classification is one of the most difficult issues in the machine learning and data mining community. In particular the problem is becoming more difficult with data sets with a large number of features many redundant features affect the efficiency of the data classification process. Specifically many biomedical data diagnosing cancer both have a large imbalance and have thousands of features. Therefore finding a solution to overcome these difficulties is extremely important and very meaningful. In this paper we present an overview of the imbalanced data classification and the difficulties encountered in current approaches from which we propose a new method SMOTE-PLS. To evaluate the effectiveness of this new method we conducted experiments based on standard cancer data sets from UCI sources including breast-p coil2000 leukemia colon-cancer and yeast. Empirical results show that the correctly classified minority samples are significantly improved which proves that the new method is more effective than the previous one in dealing with imbalanced data and the large number of features. Keywords data mining SMOTE Imbalanced data classification PLS. 1. Introduction Data classification is a widely applied problem in practice however many problems appear imbalanced data which means there is a huge difference in the number of samples of the two labels. Imbalanced data classification is one of