Using daily RON/EURO exchange rate data for the 01:2005 to 03:2013 period we test the presence of structural break using the Zivot-Andrews test and PELT algorithm. After the identification of structural breaks we generate small time series consisting of 10, 30 observations starting from the moment of the break and apply the following smoothing techniques: simple moving average, exponential moving average, a Grey model GM(1,1). We identify the best smoothing techniques using sum of squared errors (SSE) and mean relative error (MRE), in small samples GM(1,1) over-performed the moving average and the exponential moving average smoothing techniques. © Springer International Publishing Switzerland 2014.
CITATION STYLE
Scarlat, E., Zapodeanu, D., & Ioan, C. M. (2014). Comparing smoothing technique efficiency in small time series datasets after a structural break in mean. Studies in Computational Intelligence, 513, 145–153. https://doi.org/10.1007/978-3-319-01787-7_14
Mendeley helps you to discover research relevant for your work.