Unsupervised Feature Selection Technique with Enhanced Binary Bat Algorithm

Feature selection is considered as a crucial step in machine learning, particularly when dealing with dimensional dataset. Feature selection involves selecting a subset of the most relevant features from a dataset. Selection of the most relevant features from huge volume of dataset improves the performance of machine learning algorithms. Traditional approach of feature selections often relied on supervised learning techniques, which required dataset to be labeled. In contrast, Unsupervised feature selection methods, which do not require labeled data, offer a promising alternative in feature selections, it leverages on the patterns, structures and relationships within the unlabeled data. In this paper, we proposed and Enhance Unsupervised Feature Selection with Binary Bat Algorithm technique (EUFS-BBA) to improve the selection of relevant features from high dimensional dataset. Experiments were applied using four benchmark text datasets. The results show that the proposed EUFS-BBA has improved the performance of machine learning algorithm measured by F-measure and Accuracy.