search

grambank / grambank-analysed

3 stars 0 forks source link

make dual plots flexible to spatial options #69

Closed SamPassmore closed 2 years ago

SamPassmore commented 2 years ago

Hi Hedvig,

In response to #55 I made the dual plot model script a bit more flexible so that we can specify which spatial setting we like to reproduce the plot.

I did this with the idea that we can put the reproduce plots in the SM to show that the choice of spatial parameters doesn't influence the results. Actually, if anything increasing the spatial area reduces the size of spatial effects.

Currently I just have the options for the file name suffix as commented out objects, because I am not sure how you want to fit in that kind of variation (if not using the command line).

Let me know if that suits.

SamPassmore commented 2 years ago

I also updated the spatial decay parameter plot

SamPassmore commented 2 years ago

I'm not sure why R_grambank/output/coverage_plots/coverage_table.tsv changes - but this is what happens when I run the code from a fresh pull. Maybe that is expected?

The model results change slightly because there is a random element coming from the hyper.sample function and I used a different model file (the qs files with the parameter suffix rather than the ones without suffixes).

HedvigS commented 2 years ago

I'm not sure why R_grambank/output/coverage_plots/coverage_table.tsv changes - but this is what happens when I run the code from a fresh pull. Maybe that is expected?

I changed it so that some contents of output gets pushed here. This was brought on my having to pull information from PDF tables from someone else work. I don't think we should push all of output, but a few select tables is a good idea.

It shouldn't change though, that's a bit odd. i'm looking into it now. The coverage table is made by coverage_bar_plots.R which is one of several scripts that's called when running make data from the makefile. I can't think of a reason it'd change though. I'll look into it.

The model results change slightly because there is a random element coming from the hyper.sample function and I used a different model file (the qs files with the parameter suffix rather than the ones without suffixes).

That makes sense, no worries.

HedvigS commented 2 years ago

I think something is going on with git submodules and that is why the coverage table numbers are changing. i'm not able to reproduce the same numbers as Sam in my clone.

@johenglisch just to be sure, what is the git command we should run so that we have the git submodules in our local clones checkout to the same version that is on remote?

When i run git submodule update --init i get a diff between clone and remote for commit hashes

HedvigS commented 2 years ago

after the commit b9691d4 there are only 3 changed files in this branch from main. I ran the coverage barplots script on my clone.

There is also the possibility that my clone has the wrong git submodules checked out. But now at least it's the same as what I committed before. We'll figure out what's going on later.

@SamPassmore I like the changes to the dual plots script.

HedvigS commented 2 years ago

@SamPassmore can we delete this branch?