In most cases, feature sets available for machine learning algorithms require a feature engineering approach to pick the subset for optimal performance. During our link prediction research, we had observed the same challenge for features of Location Based Social Networks (LBSNs). We applied multiple reduction approaches to avoid performance issues caused by redundancy and relevance interactions between features. One of the approaches was the custom two-step method; starts with clustering features based on the proposed interaction related similarity measurement and ends with non-monotonically selecting optimal feature subset from those clusters. In this study, we applied well-known generic feature reduction algorithms together with our custom method for LBSNs to evaluate novelty and verify the contributions. Results from multiple data groups depict that our custom feature reduction approach makes higher and more stable effectivity optimizations for link prediction when compared with others.