Smoothing a dataset refers to the creation an approximating function that attempts to capture important patterns in the data while leaving out noise and other unimportant patterns. There are many forms of smoothing.