2

I have a time serie and I want to find similarities in it. For the first step I have calculated Haar-wavelet coefficients for this time serie, and now I don't know exactly how should I continue

should I extract features from this transformed data, to find similarities? how can I do that?

TangoStar
  • 419
  • 6
  • 13

1 Answers1

1

I'm not sure what you mean by "similarities" in it? You mean you want to perform a wavelet analysis or calculate the coherency with another timeseries?

To start off with a simple wavelet analysis you can use the WaveletComp package:

my.require("WaveletComp")

# read your data e.g. from netcdf file into "d1" here...

# make dataframe and wavelet 
my.data = data.frame(x = d1)
my.w = analyze.wavelet(my.data, "x",method = "white.noise",
       loess.span = 0,n.sim=10)
wt.image(my.w,color.key="interval",n.levels=200)

The package also allows you calculate coherency between 2 time series.

Adrian Tompkins
  • 113
  • 5