If I have predictions derived empirically without knowing the functional form, is there a way to calculate the prediction interval? Thanks -- WenSui Liu https://statcompute.wordpress.com/ [[alternative HTML version deleted]]