Abstract
Regression analysis plays indispensable role in QSAR/QSPR, chemical Engineering, science & technology and research projects. Best fit regression models are constantly a challenge to the researchers, efforts are taken to minimize the error components so that the predictability and efficiency of models increase. Presence of high error component eventually upset the future research and forecasting of the facts. In this paper a technique is introduced that reduces the error component and improves the predictability and efficiency of the model.