IUG Journal of Natural Studies Peer-reviewed Journal of Islamic University-Gaza ISSN 2409-4587 IUGNES Vol. 25, No 2, 2017, pp 106-116 601 IUG Journal of Natural Studies (Islamic University of Gaza) / CC BY 4.0 Accepted on (14-03-2017) 1. Introduction: The Time series analysis is used for numerous applications such as: size of population forecasting, yield projections, consumer price index prediction, census analysis, and many more. An autoregressive (AR) process is a model that future observations are predicted completely based on the past values of the time series, a moving average (MA) process is a model that future observations are predicted completely based on the past values of random disturbances, and an autoregressive-moving average (ARMA) process is a model that combines both past values of the time series and past random disturbances (Box et al., 2015). The most important objectives of time series analysis are: first, determining the characteristic of the time series by the sequence of values; secondly, forecasting future values of the time series variable. Both of these goals assume that the pattern of observed time series data is determined and formally described. Building on work on forecasting of the time series literature, we are examining the price one must pay for using ANNs under suboptimal conditions. We are testing different values of first order autoregressive parameterization, AR(1), under which relative efficiency of the ANNs model to ARIMA model, determining ranges of AR(1) coefficient,  , for which ARIMA is efficient and quantifying the effect of trend on the efficiency of the ARIMA estimator. Furthermore, we have conducted an exhaustive simulation study setup to examine the relative efficiency of ANNs to that of ARIMA models. In particular, how do ARIMA models perform in forecasting in case of linear modeling? Using ANNs and ARIMA Models to Make Accurate Forecasts for Palestinian Official Statistics Based on Simulation and Empirical Applications Samir K. Safi 1,* 1 Department of Economics & Applied Statistics, Faculty of Commerce, Islamic University of Gaza, Gaza Strip, Palestine * Corresponding author e-mail address: samirsafi@gmail.com Abstract Accuracy of forecasts of economic indicators is a major concern of statistical and economics departments. Over the past three decades there has been growing literature on applications of artificial neural networks (ANNs) to business and financial domains. ANNs do not assume restrictions during the modeling process because ANNs recognize the relationships between the variables. Thus, ANNs have the capability of executing the forecasting for different types of models without a pre- knowledge about the relationship between explanatory and response variable. In this paper, we demonstrate how ANNs can be used to make forecasts of artificial and real data sets. Furthermore, we will compare the accuracy of the ANN forecasts to those obtained by more classical time series models as autoregressive integrated moving average (ARIMA), using exhaustive simulation and real data on size of the population in the Palestinian Territories. Keywords: Population, Artificial Neural Networks, ARIMA, Forecasting, Time Series.