Microarray Data Classification Based on Neighbourhood Components Analysis Projection Method

Microarray data can contribute to precise classification of tumors. However, it is difficult to solve the problems caused by the curse of dimensionality and small sample size of the microarray data. Neighbourhood components analysis (NCA) is a method which can maximize a stochastic variant of the leave-one-out KNN score on the training data set and can effectively learn linear projection matrices for dimensionality reduction in classification. Based on NCA, a discriminant projection method, called NCADP, is proposed to solve the two problems in tumor classification task. Firstly, the project matrix is constructed by using the classification probability. Secondly, the high-dimensionality microarray data are projected into a lower dimensional space. Finally, the K-nearest neighbourhood classifier is employed to classify the tumors in the low-subspace. The proposed NCADP is non-parametric, without making assumptions about the shape of the class distributions or the boundaries between them, and it achieved higher correct classification rate than traditional ones in the experiments over two public microarray datasets, which verified that the proposed method is effective and feasible for tumor classification.