The mean of the square of the differences of the forecasts and observations For mean square error, the variable in the definition of mean is (f-x)2, where f refers to forecasts and x refers to observations The mean square error is a measure of forecast accuracy The lower the mean square error, the more accurate the forecasts