Time series analysis of building construction cost index in Türkiye

Abstract

Successful cost prediction is one of the major issues in the construction industry. Practitioners and researchers use many methodologies to determine accurate project budgets. Time-series analysis is a widely used projection determination tool, that allows accurate forecasting in many areas such as financial analysis, temperature forecast, etc. This paper aims to determine the efficiency of time-series analysis in forecasting construction costs in Türkiye. To that end, construction cost index (CCI) data between 2015-2022 were used and two main time-series analysis methods; Holt-Winters Exponential Smoothing (Holt-Winters ES) and Autoregressive Integrated Moving Average (ARIMA) have been employed. The results showed that all models underperformed in an environment with high inflation. However, considering all models, the triple exponential smoothing model showed the relatively best forecasting performance. It is suggested that the prediction performance can be improved using multivariate models and machine learning techniques. Keywords Construction cost index Time series analysis Future projection ARIMA Holt-Winters ES