In many facets of our daily lives, machine learning has been widely adopted and applied. However, as the big data era approaches, certain conventional machine learning techniques are unable to meet the demands of real-time processing for significant data quantities. Machine learning must redesign itself in reaction to massive data. In this article, we evaluate current studies that have used machine learning for big data processing. First, a review of big data is provided, and then an analysis of the new features of machine learning in relation to large data follows. Then, using machine learning methods, we suggest a workable reference framework for managing massive data. The pre-processing steps are explained in the following chapter. This research work is carried out based on three approaches, namely filter feature selection approach, hybrid approach and ensemble feature selection approach. The mentioned approaches are analyzed and the results obtained are presented.

Feature Selection Algorithms, Big Data, Machine Learning, Big Data Challenges, Machine Learning Algorithms