Efficient Multiclass Classification Using Feature Selection in High-Dimensional Datasets

Feature selection has become essential in classification problems with numerous features. This process involves removing redundant, noisy, and negatively impacting features from the dataset to enhance the classifier's performance. Some features are less useful than others or do not correlate with the system's evaluation, and their removal does not affect the system's performance. In most cases, removing features with a monotonically decreasing impact on the system's performance increases accuracy. Therefore, this research aims to propose a dimensionality reduction method using a feature selection technique to enhance accuracy. This paper proposes a novel feature-selection approach that combines filter and wrapper techniques to select optimal features using Mutual Information with the Sequential Forward Method and 10-fold cross-validation. Results show that the proposed algorithm can reduce features by more than 75% in datasets with large features and achieve a maximum accuracy of 97%. The algorithm outperforms or performs similarly to existing ones. The proposed algorithm could be a better option for classification problems with minimized features.