Previsão do preço do cimento Portland e do aço no Rio Grande do Sul por meio de modelagem wavelet- ARIMA

Abstract

This study is a price analysis of two important civil engineering commodities in Rio Grande do Sul. It is based on the examination of each series composed of monthly prices of Portland cement and steel in order to predict their prices in the state using the wavelet-ARIMA model in order to improve the sector development. Directly affected by the economic instability, the cement and steel industry decelerates. The activity level diminishes if compared to the previous month and there is less work opportunities mainly in Rio Grande do Sul. Considering this, this study focuses on the role of these materials in the civil engineering market, aiming at predicting the cement and the steel prices in Rio Grande do Sul. ARIMA methods were used which represent the stochastic process that generates the time series data whose source is the Brazilian Construction Industry. The analyzed period goes from September/1995 to December/2016, equals 256 observations of the Portland cement monthly price and 128 observations of the steel monthly price from February/2007 to September/2017. After this, the original data series is decomposed by Haar wavelets in approximations and details. The approximations are decomposed in different levels, the details are wavelets coefficients that with the scale functions reconstruct the new data series. In every discretization process there is loss. Thus, the transformed series is more accurate than the original, that is, the wavelets decomposition is used as a way to achieve a more accurate prediction when the approximation factors and the details are adjusted to the autoregressive integrated moving average. Finally, in Rio Grande do Sul, the model that best represents the steel price series is an ARIMA (1,1,0) and the model that best represents the Portland cement price series is a ARIMA (1,1,1).