Thesis (M. Sc. (Statistics)) --University of Limpopo, 2014 / analysis of price indices of electrical appliances in South Africa is performed using monthly data from Statistics South Africa for the period January 1998 to December 2010, with 2005 as a base year. Time series analysis (exponential smoothing and ARIMA) and neural networks are employed in developing forecasting models. The results for single, double and triple exponential smoothing are compared and triple exponential smoothing is found to be the best model amongst the three to forecast the electrical price indices in South Africa. ARCH models were also employed for the variable that failed to pass the requirements from ARIMA. Comparing neural networks, ARIMA and triple exponential smoothing results, neural networks is found to be the best model for forecasting price indices of electrical appliances. Regression analysis was then applied to the lighting equipment variable to check for a monthly effect after its plot depicted some seasonality pattern. Only the month of February did not have an impact or an effect on time since it was found not to be significantly different from zero. Multivariate time series is also applied in checking the correlation between the variables. Keywords: Time series analysis, ARIMA, ARCH, multiple linear regression, exponential smoothing, neural networks, electrical price indices.