How to Measure Prediction Error with STEYX

STEYX gives the typical error of a straight-line prediction, in the same units as your data — so you can quote a range, not a guess.

You will learn to

• Measure the standard error of a linear prediction
• Turn a forecast into a rough 95% range
• Compare two predictors by their error

How it works

STEYX(known_ys, known_xs) is the standard error of the estimate: roughly how far actual values typically fall from the best-fit line, in the units of y. If you predict rent from floor area and STEYX is $37, your predictions are typically off by about $37.

Use it to put a range around a forecast — about two standard errors either side covers most outcomes for points inside your data — and to compare predictors: the one with the smaller STEYX predicts more precisely.

=STEYX(C2:C11,B2:B11)