Bayesian Structural Time Series (BSTS) and ARIMA models fitted for Consumer Price Index Data in Albania

Abstract

Time series data appear in many applications in different fields such as business, physical sciences, social sciences, health, medicine etc. and making approaches to forecasting common problems are as important as the explanatory models. In this paper, we propose two models designed to work with time series data of the Consumer Price Index (CPI) in Albania monthly taken for different categories in the representative basket of goods and services. The first model, Autoregressive İntegrated Moving Average (ARIMA) is a statistical analysis model that uses time series data to either better understand the data set or to predict future trends based on past values. The second model, Bayesian Structural Time Series (BSTS) is designed to work with time series data, and it also has promising application in the field of analytical marketing. It can be used to assess how much different marketing campaigns have contributed to the change in web search volumes, product sales, brand popularity and other relevant indicators. The results from the models are taken using the programming language R and its packages for time series models. The data are taken from INSTAT (Institute of Statistics, Albania) from January 2007 to October 2023 to make predictions for the CPI index in Albania and the comparison of the results from the models too.