In this case, training days were Mondays between June 01, 2001, and July 30, 2001. Two different training sets were assembled: one for 1-interval forecast (15 minutes ahead) and other for 8-interval forecast (2 hours ahead). Testing day was taken as Monday August 6, 2001. AVG Error, MAX Error and SD for this case are 3.0%, 13.3% and 1.9% respectively. Errors in primary feeder forecast are usually larger than in transformer forecast, since daily load curves of feeders are less regular than transformer daily load curves. Fig. 7 shows reference and calculated (Cal1) curves in this case.