Using the two-population genetic algorithm with distance-based k-nearest neighbour voting classifier for high-dimensional data

Owing to developments in computer technology, high-dimensional data has become a popular research issue. However, the traditional statistical methods cannot perform well when the variable numbers (p) are greater than the sample size (n). Accordingly, this paper proposes a novel hybrid model that combines statistical methodology with data mining techniques for the classification of high-dimensional data. In the proposed model, the Fisher's least significant difference test was originally used for initial dimension reduction. Subsequently, this paper uses a two-population genetic algorithms and a non-parametric statistics classification method (distance-based k-nearest neighbour voting classifier) to evaluate and to rank the variables' importance. Furthermore, the evaluation of the relevant variables for classification is considered with the outlier detection method. Eight different public gene expression datasets are used to compare the performance of the proposed model with the existing methods. The experimental results indicate that the proposed model performs better than the existing methods in terms of the classification accuracy.