blandr module development

Everything related to the development of modules in jamovi

by jonathon » Tue Apr 17, 2018 1:23 am

hi maurizio,

with the dependencies recently added to blandr, it will need to be compiled separated for each OS.

so your experience is normal.

cheers
User avatar
jonathon
 
Posts: 869
Joined: Fri Jan 27, 2017 10:04 am

by deepankardatta » Tue Apr 17, 2018 11:03 am

I might just get rid of the options for marginal histogram, as I think the package I use to get the extra information probably will just not be helping. Go for simplicity.
deepankardatta
 
Posts: 21
Joined: Mon Jan 29, 2018 4:30 am

by jonathon » Tue Apr 17, 2018 12:46 pm

up to you. i'm happy to lend a hand getting it working if you change your mind.

cheers
User avatar
jonathon
 
Posts: 869
Joined: Fri Jan 27, 2017 10:04 am

by deepankardatta » Mon Apr 23, 2018 10:41 pm

Thanks for the offer Jonathon. As I said this is a procrastination development, and I should stop procrastinating! I'm planning to make a CRAN level release, and then park it for a little while as I need to concentrate on other things. However I may take you up on your offer later if that is OK?
deepankardatta
 
Posts: 21
Joined: Mon Jan 29, 2018 4:30 am

by jonathon » Mon Apr 23, 2018 10:51 pm

sure thing.

jonathon
User avatar
jonathon
 
Posts: 869
Joined: Fri Jan 27, 2017 10:04 am

by deepankardatta » Tue Apr 24, 2018 3:38 pm

Dear all

Here's an update version with better layout of the table (thanks all!): https://github.com/deepankardatta/bland ... blandr.jmo

There's some capitalisation issues in the table which I'll fix soon.

I suspect from your comment Jonathon, if I get rid of the ggExtra dependency for the blandr library, the Jamovi module will not need compilation for each OS?
deepankardatta
 
Posts: 21
Joined: Mon Jan 29, 2018 4:30 am

by jonathon » Wed Apr 25, 2018 6:29 am

hey! this is looking good!

so i gather bland-altman is supposed to work on continuous variables, so you should prevent the user from being able to assign 'Nominal Text' variables to the 'Method 1' and 'Method 2' options, and you can 'suggest' that people assign continuous variables, i.e.:

Code: Select all
      suggested:
        - continuous
      permitted:
        - continuous
        - nominal
        - ordinal


take a look at the dependent variable from our IS T-Test:

https://github.com/jamovi/jmv/blob/mast ... ml#L38-L46

I suspect from your comment Jonathon, if I get rid of the ggExtra dependency for the blandr library, the Jamovi module will not need compilation for each OS?


yes i think so :)

this is all looking good!

jonathon
User avatar
jonathon
 
Posts: 869
Joined: Fri Jan 27, 2017 10:04 am

by deepankardatta » Wed Apr 25, 2018 5:01 pm

Thanks! I've limited the data to continuous variables only as that's what it's supposed to work on as a method comparison tool. Have added it in: https://github.com/deepankardatta/bland ... blandr.jmo

Is there a way to link the Jamovi module to a help website (e.g. a GitHub wiki)? Also would you expect the JMO file to be c. 13MB? I've slimmed down the blandr package for CRAN to <500kB by excluding the "build" folder from the building process - if I got rid of the inst/docs folder I could get it down to <50kB.

How do you think the module is for submission to the Jamovi repository? I think I have most of the things I need to make it useful. (Although the WALRUS module made me want to design a custom icon for the module in the future.)

I'm hoping the package can be fixed soon for both a Jamovi library and CRAN release. I've done some other changes to get it CRAN ready - the only 2 notes are (1) Jamovi is a non-standard directory, and easily explained to the CRAN team (hopefully), and (2) a ggplot2 related issue which I think I've solved.

Thanks again for everyone's help!
deepankardatta
 
Posts: 21
Joined: Mon Jan 29, 2018 4:30 am

by jonathon » Wed Apr 25, 2018 11:25 pm

hey,

so wrt to only allowing continuous variables, our policy is to allow nominal, ordinal and continuous, when an analysis calls for continuous variables. this is because we only use the variable type as a guide. a lot of people just load up a data set and get analysing, and don't worry about assigning all the measure types correctly. if your data set has a lot of columns, setting them all up correctly can get tedious.

a 13 MB .jmo file is fine.

wrt linking to a website, we haven't done a great job on this front. what you can do is include a link in the description in the jamovi library.

wrt a CRAN submission, you should add the `jamovi` folder to your .Rbuildignore file, and then it will be (correctly) excluded from the CRAN submission.

wrt adding it to the jamovi library, if you could write a little blog post on blandr, that we could post on blog.jamovi.org that would be great :) a good way to do it, would be to write two blog posts, basically the same, but with some things in a different order.

one that says "blandr for jamovi is here ... this is what it does ... oh, and it's also available for R", and another that says "blandr for R is here ... this is what it does ... oh and it's also available for jamovi".

we'll post them in such a way that the first is listed in our blog, and the other will be a little bit hidden, but will get picked up by r-bloggers. that way you're reaching out to both communities.

jonathon
User avatar
jonathon
 
Posts: 869
Joined: Fri Jan 27, 2017 10:04 am

by deepankardatta » Mon May 07, 2018 3:57 pm

Hi Jonathon

Have sent you a message via the contact (at) email address.

With Thanks
Deepankar
deepankardatta
 
Posts: 21
Joined: Mon Jan 29, 2018 4:30 am

PreviousNext

Return to Module development