This paper presents a new methodology for house-hold electricity demand forecast using a hybrid method combining nonparametric model and time series analysis. The relationship between outdoor temperature and electricity demand is studied in order to develop an approach to modeling and analyzing the electricity demand in response to temperature changes. First, the kernel density estimation as a nonparametric method is used to examine the mentioned relationship and provide probability distribution of future possible demand values. Second, two autoregressive (AR) models are applied to forecast total power demand using different information sources. These sources comprise results of forecasting temperature/electricity demand relationship as well as nonparametric residual data. The performance of the methodology represented in this work is evaluated using a comparison to the results of an ARMAX model. The forecasting accuracy of the models is compared on the basis of mean absolute error (MAE) and mean absolute percentage error (MAPE) metrics. The forecasting results demonstrate that the hybrid model performs remarkably well and thus is more favorable than ARMAX model. This study employs real data for numerical analysis of proposed methods to increase the results efficacy.