A filter feature selection for high-dimensional data

In a classification problem, before building prediction model, it is very important to identify informative features rather than using tens or thousands which may penalize some learning methods and increase the risk of over-fitting. To overcome these problems, best solution use feature selection. this article, we propose new filter method for selection, by combining Relief algorithm multi-criteria decision-making called TOPSIS (Technique Order Preference Similarity Ideal Solution), modeled selection task as decision problem. Exploiting methodology, matrix computed delivered Technique Solution in order rank features. The proposed ends up giving ranking from mediocre. evaluate performances suggested approach, simulation study including set experiments case studies was conducted on three synthetic dataset scenarios. Finally, obtained results approve effectiveness our detect