add prior check figure

[SamPassmore]

@HedvigS

I built a graph in the same vein as the spatial parameter plot to show the priors have little effect on the estimates. I'll leave it to you to decide if the plot goes in the SM or not, but at least we can be safe in the knowledge that the priors do not matter. Currently, we just write the priors have no effect.

[SamPassmore]

whoops - forgot to change the y-label. I have done that now!

[HedvigS]

Thanks Sam!

I'm going to revert one of your commits because it changes some extra stuff but I'll put back in the new script.

[HedvigS]

I re-ran the spatial parameters effects plots. It looks a bit different from the one in the ms google doc, but still overall the same story. This could be to some random differences when trees are pruned etc for analysis. Does it look okay to you @SamPassmore ?

[SamPassmore]

Sorry - I thought I didn't commit those changes, but obviously not! Thanks for reverting them.

I think it is probably worth investigating why the plot changed, even though the overall tale is not that different. But I think our output is out of sync. I see you changed the regex to "GB[0-9]{3}kappa._pcprior0.1..qs", which is sensible given the change in the output naming style.

But in the output link #45 the only output that ends with pcprior always has the same spatial settings (kappa = 2, sigma = 1.15), so I can't investigate further, sorry. I can see model output with different spatial settings (e..g in the form GB083_kappa_2_sigma_2.RD), but these don't end with pcprior, which leads me to think we are looking at different outputs.

[HedvigS]

Sorry - I thought I didn't commit those changes, but obviously not! Thanks for reverting them.

I think it is probably worth investigating why the plot changed, even though the overall tale is not that different. But I think our output is out of sync. I see you changed the regex to "GB[0-9]{3}kappa._pcprior0.1..qs", which is sensible given the change in the output naming style.

But in the output link #45 the only output that ends with pcprior always has the same spatial settings (kappa = 2, sigma = 1.15), so I can't investigate further, sorry. I can see model output with different spatial settings (e..g in the form GB083_kappa_2_sigma_2.RD), but these don't end with pcprior, which leads me to think we are looking at different outputs.

Yes, we may be looking at slightly different outputs. If we agree on the scripts in principle, what I'll do is I'll delete all output, merge in this branch and then re-run again. Sound good?

[SamPassmore]

i would think that we should have all the output somewhere. I assume that you have the other outputs somewhere - could you up load those to the shared folder? Or would you rather burn and re-build?

[HedvigS]

i would think that we should have all the output somewhere. I assume that you have the other outputs somewhere - could you up load those to the shared folder? Or would you rather burn and re-build?

Not sure what you mean exactly, I have put the output in the nextcloud.

I'll burn and rebuild anyway :)

[SamPassmore]

I think we just have our wires crossed. So, we do have files in nextcloud that contain all the different spatial parameters, but these files do not end with "pcprior". The regex in 2cec44c was changed to have pcprior on the end. This made me think that the output in nextcloud that I need to analyse should have pcprior on the end, but the only files that have the pcprior suffix have the spatial settings kappa =2 sigma = 1.15 - so I though that meant there was other output that hadn't been uploaded.

[SamPassmore]

If we're starting again though it doesn't matter.

[HedvigS]

There are files in the nextcloud that specify the priors. There are also files that don't, I've been continuously adding files there so there are also older files.