An Improved Artificial Bee Colony for Feature Selection in QSAR

Quantitative Structure–Activity Relationship (QSAR) aims to correlate molecular structure properties with corresponding bioactivity. Chance correlations and multicollinearity are two major problems often encountered when generating QSAR models. Feature selection can significantly improve the accuracy interpretability of by removing redundant or irrelevant descriptors. An artificial bee colony algorithm (ABC) that mimics foraging behaviors honey was originally proposed for continuous optimization problems. It has been applied feature classification but seldom regression analysis prediction. In this paper, a binary ABC is used select features (molecular descriptors) in QSAR. Furthermore, we propose an improved ABC-based QSAR, namely ABC-PLS-1. Crossover mutation operators introduced employed onlooker phase modify several dimensions each solution, which not only saves process converting values into discrete values, also reduces computational resources. addition, novel greedy strategy selects subsets higher fewer helps converge fast. Three datasets evaluation algorithm. Experimental results show ABC-PLS-1 outperforms PSO-PLS, WS-PSO-PLS, BFDE-PLS accuracy, root mean square error, number selected features. Moreover, study whether implement scout tracking drawing such interesting conclusion dealing low-dimensional medium-dimensional