phase-randomized correlation values output

[arfriedman]

Greetings,

I just started using pyleoclim -- thanks for this excellent package.

I am trying to construct a confidence interval for the correlation statistic (r) using phased-randomized correlation.
Is there a way to obtain the significant value of r (or the full distribution of r values) from the correlation output?

Best,
Andrew

[CommonClimate]

Hi Andrew, glad you are enjoying the package. As explained here, Pyleoclim's Corr objects contain :

• r (float), the correlation statistic
• p (float), the p-value, which is used to decide whether the correlation is in any way remarkable (“significant”)
• signif, a boolean indicating whether significance was reached, at the indicated level (0.05 by default)

We do not currently export the distribution of surrogate r values, but I suppose we could. However, doesn't signif or p give you what you want?

[arfriedman]

Thank you for your reply @CommonClimate.

In addition to the Corr object contents, I would be interested in using the distribution of surrogate r values to visually represent how anomalous the calculated r value is. For example, see this paper (Figures 1f, 2c-d, and 2h). [I previously made these in Matlab].

If I understand correctly, could I do this using the values rSim from corr_isospec in correlation.py?

[CommonClimate]

Yes, rSim is that variable but:

1. it is not currently exported by corr_isospec
2. the latest version of the package (1.1 and later) no longer uses this function. Instead, everything is done through the SurrogateSeries class and the correlation.py:association() function.

You can either make your own version of corr_isospec, which would do the job for this (if all you care about is Pearson's correlation, and not a wider set of metrics) or open an issue to ask us to export the simulated values of r in the correlation object. @khider will likely have objections because this is a change to the API, but since we're adding a field and not renaming/removing, I don't anticipate that impacting backward-compatibility, so it should be OK.

PS: please say bonjour to Gilles, Myriam and Guillaume

[arfriedman]

Thanks @CommonClimate. I will pursue both these suggestions.
PS. Merci, will do!

[CommonClimate]

addressed by #634