Determining the Best Value of Window Length (L) in Singular Spectrum Analysis (SSA) Method

Authors

  • Nurul Jannah Amdayania
  • Gumgum Darmawan
  • Yudhie Andriyana
  • Neneng Sunengsih

Abstract

Singular Spectrum Analysis (SSA) is a forecasting method that involve trajectory matrix to decompose
the time series data. Trajectory matrix is a multidimensional series as a result of transformation of unidimensional
time series data. The application of trajectory matrix consists a window length (L) as a dimension of its matrix.
The calculation to obtain L value is based on autocorrelation function of monthly rainfall data in the nearby area
from Juanda Airport, Surabaya during the period of January 1981 until December 2012. The study produces L
value in amount of 107 and indicates that the calculation to obtain L value based on autocorrelation function is
more reliable than the trial and error process.
Keywords: Autocorrelation, Blind Source Separation, Rainfall, Singular Spectrum Analysis, Window Length.

Downloads

Published

2017-04-01

Issue

The 2nd International Conference On Applied Statistics (ICAS 2016), Departement of Statistics, Faculty of Mathematics and Natural Sciences, Universitas Padjadjaran. ISSN: 2579-4361

Section

Articles