An Adaptive Dandelion Optimization Algorithm for High Dimensional Big data Feature Selection Problems

Feature Selection is an essential process that helps in removing the insignificant features from the data. The process of feature selection is a challenging issue due to the combinatorial nature of the features. The researchers have focused on utilizing meta heuristic-based feature selection approaches which are robust for high dimensional datasets. By considering this, this research developed an Adaptive Dandelion Optimization (ADO) to select the important feature sets. The features from the high dimensional data are minimized using Principal Component Analysis (PCA) and the proposed ADO performs effective feature selection using the adaptive tent chaos mapping strategy. The adaptive tent chaos mapping strategy uniformly distributes the initial population with a qualitied initial population. The tent chaos mapping strategy adopts the initial values in a randomized sequence that helps in the effective selection of features. The PCA-ADO achieved better classification accuracies of 98.50% and 99,12% for two difficult classes such as 9 Tumor and 11 tumor datasets respectively.