smoething : Java Glossary

Smoothing is partly a cosmetic process, to show graphs with noise removed so the general trends are more obvious. You don’t end up with a mathematical approximation, as you do with curve fitting, just nice looking data. The classic technique for curve fitting, polynomial approximation, does not work well for smoothing. When you increase the order of the polynomial to capture the various humps in the data, the approximation shoots out wildly between the anchoring data points. There are special techniques for smoothing. Moving average is one of the simplest to implement. For example you would display your weight averaged over the last week. Each day, the oldest day drops out of the average and today takes its place. This has the added advantage of ironing out weekly cycles. It also irons out daily fluctuations due to salt intake and bowel movement.