Raqamli Transformatsiya va Sun’iy Intellekt
COMPARATIVE ANALYZING FEATURES SELECTION METHODS FOR DATA MINING TASKS
1 Tashkent University of Information Technologies named after Muhammad al-Khwarizmi.
Abstract
This paper discusses feature selection methods for data mining and machine learning, focusing on three main approaches: wrapper, filter, and hybrid methods. These techniques help reduce dimensionality, improve computational efficiency, and enhance model accuracy by selecting the most relevant features and eliminating unnecessary data. Additionally, the paper presents a software tool designed to facilitate the feature selection process, utilizing the Java Data Mining API for efficient and scalable implementation. The software allows users to process large datasets and apply different feature selection techniques based on specific requirements. The paper also outlines the steps involved in the feature selection process, providing insights into its practical application. By combining a review of feature selection methods with a practical software solution, this study aims to assist researchers and practitioners in selecting the most suitable techniques for data preprocessing in machine learning. The findings contribute to improving model performance and optimizing computational resources, making machine learning applications more effective and efficient.
References
No references listed for this article.