The aim of this paper is to evaluate the performance of median filter to extract accurate mathematical model of household electric power load. The model is expressed as a function of voltage and phase angle. The model is expressed by first order differential equation. The median filter is used for smoothing the raw data which may contain noise. Filtered data with low frequency sampling is then utilized to estimate model parameters. In this paper the advantage of Least Square Method is explored for finding the true value of model parameters. The effect of different median filter size on model accuracy is analyzed. The obtained model can be developed further for advancing Non-Intrusive Load Monitoring (NILM) technique. In this paper, four sizes of median filter have been evaluated. Simulation result has shown that different size of median filter resulted different parameters. Simulation has also shown that among four size of median filter, median filter with 5-window presented more accurate model proved by lower Root Mean Squared Error.