Study on Quantitative Structure-Retention Relationships (QSRR) for Oxygen-Containing Organic Compounds Based on Gene Expression Programming (GEP)

Gene Expression Programming (GEP) is a novel genetic algorithm, a highly effective, stable random searching method. We take GEP to make models of Quantitative Structure-Retention Relationship (QSRR) for a series of oxygen-containing organic compounds of GC retention index, and compare the predictive results with Artificial Neural Network (ANN) and Multiple Linear Regression (MLR). The correlation coefficient on OV-1 column is 0.9919, 0.9891 and 0.9911 for GEP, ANN and MLR respectively, on SE-54 column is 0.9955, 0.9892, and 0.9917. It is shown that the predicted results by GEP are in good agreement with experimental ones, better than those of ANN and MLR.