search

Armand1 / Evolution-Revolutions

This is a continuation of the Evolution and Ecology text mining project
1 stars 0 forks source link

FN Plot #3

Open SamMckaylin opened 4 years ago

SamMckaylin commented 4 years ago

None of the FN plots have produced the Blue and Red overlays which would signify revolutions or conserved periods (as seen in "On Revolutions". In every test I've performed, at least one revolution has been detected and outputted in the table and the plot of "R". The colour palette exists and I haven't changed any code associated with the FN plot.

Could it be that the cut off for the object res2 which stores revolutions is too stringent? If so, why are they still detected in the other plots?

I've attached the plots for reference: "R"plot - Science FNplot - Science

Armand1 commented 4 years ago

So, I have to say that I don't think you should be looking for revolutions in single journals. You can -- but I suspect that the data are too thin to do so. Though you're not absolutely wrong to try. I mean Evolution (the journal) would, for example, be interesting.

I am not quite sure that I see the problem. If you don't see red, then there ain't any time points that show revolutions. But, in fact, you must get more time points that show revolutions than those expected by chance (bear in mind that you are doing lots of tests). Be prepared to run the analyses in front of me tomorrow, and we'll look at the output together.

SamMckaylin commented 4 years ago

My intention was to see whether the trends seen across a topic_discipline are replicated within journals. I've looked at Evolution, American Naturalist and Science so far and I'm trying to get PNAS to work, although it's ended up being quite finicky.

I'll bring everything along tomorrow and can show you what seems to be going wrong / discuss whether it's an approach worth taking.

SamMckaylin commented 4 years ago

Tweaking the code with that from On Revolutions has resolved the issue and revolutions + conserved periods are now coloured: Screenshot 2020-03-16 at 11 06 41

I haven't differenced the data yet as I need to run the stan code first but everything is now working as planned.

Armand1 commented 4 years ago

hi Sam --- sorry been busy with MScs. yes, that looks much more like it. I take it that is not on differenced data? And what subset of the data is that? Whatever it is, you'll find a revolution in the 70s onwards, but I would not claim anything earlier on.

I think that, in general, we'll rewrite the history of ecology/evoutonary biology. It's all happening in the 60s onwards. New subfields (ecological evolutionary stuff) and the rise of statistics I think.

SamMckaylin commented 4 years ago

No worries, it was every paper and wasn't differenced. Just wanted to find if it worked. I've got the STAN stuff and will try to get it to work and get rho values.

Everything I've seen so far seems to start in the mid 60s, I'll update once it's all differenced