This paper will try to give a clear understanding of generalized additive models (GAM) and the relationship between the response variable and the predictors when linear regression is not appropriate.