There are two crucial questions to consider when using a fitted curve to interpolate or extrapolate unknown values:

• How good is the fit? A perfect fit would mean a curve that passed right through every data point. This may not be what we want, however – especially if the data contains noise or experimental errors that we'd like to ignore while concentrating on the general trend. A good fit usually follows the trend in the data by passing a little bit above some of the data values and a little bit below some of the others, for a "net" fit that looks "just so" (as Goldilocks might say).

• Does the trend continue outside of the data? Even a curve that closely fits the data doesn't guarantee accuracy when interpolating and extrapolating outside of the data. An equation for a fitted curve may describe a general trend in the data, but that doesn't mean that the trend will be followed between, or beyond, the given data. For all we know, values could fluctuate wildly between data points, or they could take a sharp curve in a new direction.