apf_jingrui_li

Note that since the actual, measured value is greater than the predicted value of y for
2004.1, the error term is positive. Had the actual, measured value been less than the
predicted value, the error term would have been negative.

This is how they say it, I see no mistake here.