Relative humidity (RH) monitoring is of great importance for the management of water resources, agricultural field, climatic studies, as well as for public health management. The objective of this paper is to analyze, model, and forecast the monthly RH at Brasília city, Distrito Federal, Brazil. RH is given in percentage terms, i.e., assume continuous values in the interval (0,1). In this situation the traditional time series models, like the ARIMA class, are not suitable. Thus, the use of beta autoregressive moving average models (βARMA) is required. This model proposed by Rocha and Cribari-Neto was developed for rates and proportions variable with beta distribution. For this study, the βARMA model was implemented in R language. Its application to RH data was adequate, capturing the behavior of the RH series and producing accurate predictions.
βARMA; forecast; relative humidity; time series
