Milestone 06

[dy-lin]

Milestone 06

• Apply TA feedback on your dashboard, fix any bugs and implement new features (issue CM111- In-Class Feedback #50)
  • Clean up your repository (including issues) and make it ready for public release
• Finalize your dashboard
• Deploy your app on Heroku

Task 1: Apply TA feedback (30%)

Your TA will be giving you feedback on your Dash app by Thursday at 6 PM. That should be plenty of time to fix any bugs that exist, implement new features, and improve the layout of your app. You should take this opportunity to clean up your repository and make it ready for public release. Here are some suggestions on things you can do to “clean up your repository”:

• Remove .Rproj from your repo (if you accidentally committed it on github) and add it to your .gitignore file
• Close stale issues
• If your folder structure is messy or confusing, clean it up and re-organize it
• Create a v1.0 tagged release !
• You can split up your app into multiple files; just a warning, this can get complex so only do this if you want to challenge yourself and everything else is complete! Here are some instructions on how to do this.

Task 2: Finalize your dashboard (10%)

• Make sure your app is complete, functions correctly, and does what you want it to. One (optional) way to make sure of this, is to get others to run your app and give you feedback.

Task 3: Deploy your dashboard (50%)

As shown in cm111, you should deploy your app on heroku so that it is publicly accessible.

• In your README, please include the Heroku app link and clearly mark it so that we can find it!

As part of this task, in addition to a correctly deployed app, we will also check the following files:

• Procfile
• app.json
• heroku.yml
• init.R
• Dockerfile

Note: depending on how many dependencies your app has, this can take quite a while to get right, don’t leave it to the last minute!

That’s it! You have now created an analysis pipeline and deployed a dashboard app from scratch! You should all be super proud of the work you’ve done over the past 6 weeks!

[dy-lin]

Hi @Nima-Jamshidi, according to this issue, it doesn't make sense to deploy the dashboard before we get the feedback, but I will start writing the files init.R file for deployment.

After implementing feedback, I think we can separate our app.R into the dash_components.R and dash_functions.R like we did in cm110.

I think our repo is pretty clean as it is.

Some things I am thinking of implementing if we have time:

• Redoing the corrplot component using ggplot as I did for the dashboard
• Reknitting the HTML and the PDF of the final report (i.e. milestone3)
• Issue Future work #41 regarding reorganizing the data directory to be more intuitive -- this will be a larger undertaking as we would have to change hardcoded locations for each rds in our Rscripts -- I personally don't agree this this piece of feedback and think the way our data folder is, is absolutely fine

[dy-lin]

@Nima-Jamshidi, we have received feedback from Yulia #54, and I will implement those changes today. Still waiting on Firas #50.

[dy-lin]

Yulia said that her feedback were more suggestions, so I implemented the ones that I agreed with and closed the issue. I'll create the pull request and merge soon!

[dy-lin]

@Nima-Jamshidi, since I was the one to implement Yulia's feedback / it was feedback for my portion of the dashboard, to evenly distribute the workload, you can move parts of our app.R script to dash_components.R and dash_functions.R, if you have time.

Please also check over init.R (see if I am missing any packages maybe?) and the other deployment files (I vaguely remember Firas saying the init.R is probably the only file we will edit, but check anyways). Then you can tag a v1.0 release, and we can try deploying the dashboard on Heroku.

Since our app has a lot of dependencies, it would be wise for us to attempt deployment early in the week.

Note: depending on how many dependencies your app has, this can take quite a while to get right, don’t leave it to the last minute!

[dy-lin]

@Nima-Jamshidi, let me know if you do not have time to work on this milestone-- I am worried about doing this one last minute due to possible deployment issues. I have the time to do it, but I also don't want the work distribution to be uneven unless you are really busy.

[dy-lin]

Also, let me know if you have an issues in deployment. I know that you're on a Windows computer and that might make things more complicated.

[dy-lin]

Hi @Nima-Jamshidi,

according to some issues (#82) after splitting the app.R into multiple files for deployment, I think it's best if we just skip that step since it was optional anyways.

I have tested the deployment with my fork, and looks like it deployed fine.

Please deploy the app using the group repo (i.e. NOT your fork) and name it stat547-group01-2020 if the name is not taken.

I am updating the README to include the link stat547-group01-2020.herokuapp.com as where to access our dashboard.

[dy-lin]

Hi @Nima-Jamshidi since it can take a long time to deploy, I'll tag the release and deploy by 3pm tomorrow April 7 (since it's due April 7 6pm) if I haven't heard from you then.

[Nima-Jamshidi]

Hi @Nima-Jamshidi, according to this issue, it doesn't make sense to deploy the dashboard before we get the feedback, but I will start writing the files init.R file for deployment.

After implementing feedback, I think we can separate our app.R into the dash_components.R and dash_functions.R like we did in cm110.

I think our repo is pretty clean as it is.

Some things I am thinking of implementing if we have time:

• Redoing the corrplot component using ggplot as I did for the dashboard
• Reknitting the HTML and the PDF of the final report (i.e. milestone3)
• Issue Future work #41 regarding reorganizing the data directory to be more intuitive -- this will be a larger undertaking as we would have to change hardcoded locations for each rds in our Rscripts -- I personally don't agree this this piece of feedback and think the way our data folder is, is absolutely fine

I did the first two parts. I also believe that we do not need reorganization.

[Nima-Jamshidi]

I started to deploy the app without the component and function R scripts. I hope it works!
Sorry for my late response, I'm too deep into other projects and presentations.
Cheers!

[dy-lin]

Looks like it works! Awesome!