Leukemia Classification using Principal Component Analysis and Ensemble Learning on Gene Expression Data

Leukemia, a group of blood cancers, poses significant challenges in accurate diagnosis and treatment. The conventional cancer classification methods are reported to have limitations. Systematic methods based on global gene expression analysis have been presented in order to better understand the problem of cancer classification. In this paper, we propose a machine learning model for leukemia classification based on dimensional gene expression data. The model uses Principal Component Analysis (PCA) feature selection method and ensemble learning Stacking algorithm. The proposed model results in 95.5% accuracy and precision, recall and fl score of 0.96, 0.95 and 0.95, respectively, which outperforms other traditional machine learning algorithms when training on same used dataset. This result is good when solving multiclass classification problem compared to other binary classification research papers using gene expression data. In addition, we came into a conclusion that ensemble learning outperforms other methods when dealing with high-dimensional and small-sampled dataset.