Rank: 1892 based on 66 downloads (last 30 days) and 6 files submitted
photo

Agne Burauskaite-Harju

E-mail
Company/University
Linköping University

Personal Profile:

 

Watch this Author's files

 

Files Posted by Agne Burauskaite-Harju View all
Updated   File Tags Downloads
(last 30 days)
Comments Rating
14 Oct 2011 Time-lagged assymptotic dependence Tail dependence coefficient and extremal variogram with time lag, summarizing parameters Author: Agne Burauskaite-Harju statistics 7 0
14 Oct 2011 Time-laged association measures Cosime cross-similarity and summarizing parameters Author: Agne Burauskaite-Harju statistics 8 0
19 Apr 2011 Trend-Duration-Frequency Test for trends in extreme values of data for several quantiles and data resolutions Author: Agne Burauskaite-Harju statistics, mathematics, signal processing, finance, data exploration, earth science 28 1
  • 3.0
3.0 | 1 rating
19 Apr 2011 Screenshot Spatial correlation in high resolution climate data Characterising spatial patterns of correlation in high resolution climate data Author: Agne Burauskaite-Harju statistics, mathematics, data exploration, earth science 12 0
19 Apr 2011 Window cross-similarity Time series window cross-similarity Author: Agne Burauskaite-Harju statistics, mathematics, signal processing, finance 3 0
Comments and Ratings on Agne Burauskaite-Harju's Files
Updated File Comment by Comments Rating
23 Oct 2012 Trend-Duration-Frequency Test for trends in extreme values of data for several quantiles and data resolutions Author: Agne Burauskaite-Harju Patrick

Tried to run the TDF function. The description is rather short and an example is missing (at least to my feeling). Finally, I had a problem with the decluster function, which is possibly due to incorrect setting of the variable resolution (in the TDF function it is said:

% resolution - a vector of multipliers. Default - 1.
%
% For example, if obs_per_day=24 resolution=[1 3 12]results will be
% calculated for hourly observations, for a sum of 3 hour observations and
% for a sum of 12 hour observations. Note that the lowest resolution that
% procedure can handle is one observation per 3 days.

I created a vector of ones for the resolution, and used daily data. The resulting warning is:

Decluster parameter too large

Please clarify how to set resolution e.g. for a time series with daily resolution.

Contact us