Group-Aware Recommendation using Random Forest Classification for Sparsity Problem

Objectives: However, data sparsity is the most challenging issue in collaborative filtering, which arises due sparse rating matrix. To overcome this issue, we propose a group recommendation, which divides a larger task into smaller tasks to the subgroups. Many existing approaches considered only isolated subgroups. Methods: In this work, we proposed top-N recommendation algorithm by considering the interrelationship between each group which leads to an efficient and accurate way to recommend items. Findings: The proposed work is based on matrix factorization to predict the missed values in the rating matrix and exclusively constructs approximation matrix to increase the prediction accuracy using the coordinate system transfer method. The experimental results of our approaches perform better than the traditional recommendation approaches. Applications: In this method used in Movie recommendations, Product Recommendations, Travel Recommendations etc.