Using Time Series for Predict the Standard Numbers for Residential Rent in Iraq for the Years 2018-2021

Abstract

The technique of analyzing time series is considered to be one of the important statistical methods used in the field of prediction, studying a certain phenomenon in a certain period of time allows us to identify its pattern and use it to predict the future values of this phenomenon which helps setting out the right plans, The research aims to choose the best model to predict the standard number of rent prices of houses in Iraq, by using the samples of Box-Jenkins, exponential smoothing, and then using these models to predict and choosing the best one among them comparing the models by using general trend criteria (MAPE and MAD) for comparison. The models put by scientists Box & Jenkins have been relied on by setting a methodology in studying and analyzing models of the ARIMA of time series beginning from the process of stability test through auto correlation and partial auto correlation which showed instability in average as it has a general trend and stabilizes after taking the first difference by relying on Augmented Dickey-Fuller Test (ADF) to check stability and the best model has been identified by relying on criteria of used accuracy (Akaike Information Criterion AIC, Hanan-Queen Criterion (H-Q) and Schwartz-Albizi Criterion (SMC) then great possibility function has been used in evaluating model parameters in which sample ARIMA(0,1,1) has been chosen as the best model of Box-Jenkins and assets of the chosen model were tested through uni-reliability test and Ljung – Box test of the investigated phenomenon, The comparison between the best model of Box-Jenkins and exponential smoothing methods have been considered and the model of duplicate exponential smoothing (Holt's method) has been chosen as the best model of exponential smoothing relying on statistical criteria (MAPE, MAD and MSD) and model assets have tested and inspection of model efficiency through uni-reliability (auto correlation and partial auto correlation), By comparing predictions it has been concluded that prediction by using the model of duplicate exponential smoothing gives better results than using Box-Jenkins model according to prediction accuracy criteria (MAPE, and MAD) and through estimations there appears to be an increase in the standard number of rent prices of houses in Iraq, at 2016-2021 year predictive , the Data of the standard number have been collected from the Ministry of Planning/Central Agency of Statistics/Directorate of Standard Numbers (from January 2008 until December 2015) as statistical programs gretl and Minitab have been used to analyze the time series