A residential load forecast must account for weather, household behavior and changing appliance use. This Mosul case study shows how a multi-layer perceptron used disaggregated load components and temperature data to predict daily peak demand one week ahead.