alice-i-cecile / dend.ro Goto Github PK

After running FES, the fit statistics remain displayed after running an RCS/SFS analysis.

Use the $settings attribute :)

Suffers from 3-effect pathology.

Change to FES::GAM::Auto?

Only happens when FES -> GAM is selected before switching.

I can't seem to see how to pass normal vs. log-normal noise in the synthetic data functions "crude_synth_TRA" and "modern_TRA". For example, in crude it looks like lognormal is hardcoded, or maybe I don't know what I'm looking at.

These should be listed as NA (rather than NaN.NaN) or hidden. The following statistics are missing:

FES::Linear: None

FES::Nonlinear: R^2, Adj. R^2

FES::GAM: R^2, Adj. R^2 (I thought these were computed)

RCS: All (properly hidden)

SFS: All (properly hidden)

These are the average values, so it's meaningful whether a point is above or below this line.

Can it handle the http base pages (http://hurricane.ncdc.noaa.gov/pls/paleox/f?p=519:1:6550064172957880::::P1_STUDY_ID:13752) as well as the direct FTP links (ftp://ftp.ncdc.noaa.gov/pub/data/paleo/treering/measurements/europe/brit063.rwl)?

Send me a working link and I'll try it out.

Beside the download links?

A help bubble would be nice too.

There seems to be a 3 line header at the top of each file:

LASF 1 Le Communal de la Sagne PCAB
LASF 2 Switzerland Norway Spruce 1180M 4703 00648 __ 1853 1995
LASF 3 O.U. Bräker A. Zingg
LASF10111882 1053 901 925 509 510 473 533 604

Should be listed as \sigma^2 instead.

Error bars can be generated for all FES outputs.

Will need to update UI to match. Should be consistent with analysis tab; no need to let them mix and match. Live growth model at the top of this screen would be a big asset.

Remove I if \sigma_I = 0
Remove T if \sigma_T = 0
Remove A if trend is flat and m=0, b=0/1 (depending on form)

Breaks multiplicative models and is probably noise in additive ones. These need to be censored.

Always appear to be 1/1/0 respectively, even when other goodness of fit stats change appropriately.