Forecasting data using box-jenkins procedure: a case study for unemployed people in Malaysia

In forecasting and time series analysis, the Box-Jenkins approach has been widely employed as it is among the most effective methods to forecast business and non-business data. This paper focuses on modelling and applying the Box-Jenkins model to forecast how many unemployed people there are in Malaysia. Hence, the time series data of unemployed people in Malaysia between January 2010 and November 2022 were used for this study’s purpose. There are three main stages in constructing the best model: model estimation and validation, identification, as well as model application. In the first stage, the actual data on unemployed persons in Malaysia was found to be non-stationary, with a rising trend pattern. As a result, the data series was subjected to the Box-Jenkins model with first-order differencing. Hence, three models were exhibited, which are Autoregressive Integrated Moving Average (ARIMA) (1, 1, 0), ARIMA (2, 1, 0) as well as ARIMA (3, 1, 0) by referring to both time plot and Autocorrelation Function (ACF) plots. The most appropriate model was determined in the second stage, which is ARIMA (3, 1, 0) by comparing the Bayesian Information Criterion (BIC), Root Mean Square Error (RMSE) as well as Mean Absolute Percentage Error (MAPE). Finally, the best model was applied to forecast unemployment, and an increasing trend in the number of unemployed was estimated to occur for the following year.