Classifying Acute Lymphoblastic Leukemia (ALL) Microarray Samples Using K-Nearest Neighbor

Microarray experiments are used to generate gene expression data to analyze large number of problems. Human cancer is one of the serious problems and microarray analysis also got acceptance for diagnosis and classification of human cancer. But classification of microarray sample is also tricky. The problem of classifying microarray samples into a set of alternative classes is discussed here. Patients are classified into a pre-defined set of genetic mutations of acute lymphoblastic leukemia. Acute lymphoblastic leukemia (ALL) is a type of cancer which is genetic in nature. It is a heterogeneous disease that contains various leukemia subtypes that differ in their response towards chemotherapy. In this paper DNA microarray is used to detect the number of patients suffering from cancer by using ALL dataset. The dataset used here has characteristics of feature selection i.e. how to reduce the number of features that describe each observation. It contains data set of 128 patients at beginning with 12625 genes or feature and at end 94 patients are detected positive for cancer with only 743 genes or features left.