From black_holes_backend created by amjcosta : codeforgoodconf/black_holes_backend#4

We may need to add Gaussian smoothing to the processed_data. @rlbyrne had worked on this with previous data processing.

From black_holes_frontend created by sauln : codeforgoodconf/black_holes_frontend#6

Add pre-commit hooks to the git repo so that when a user pushes their changes to master, it will be rejected (maybe?) if the tests do not pass.

• Run all unit tests
• Run Flake8
• Examine code coverage

From black_holes_backend created by sauln : codeforgoodconf/black_holes_backend#7

Currently, all the data is stored directly in the git repo. As we get more data to train on, this will become too big. The data should be stored somewhere that can be retrieved via a wget command.

• Consolidate raw data in a single zip
• Store the data somewhere (George Mason servers? AWS free-tier?)
• Develop setup script to pull data from serve and run basic preprocessing.

From black_holes_frontend created by edbreeden : codeforgoodconf/black_holes_frontend#8

I'm pretty sure that right now the humans input True if there IS a WR bump [meaning NOT a blackhole] and the machine marks galaxies as True if there is NO WR bump [meaning it IS a blackhole]. lets please make sure it does not do this before we use the app to train the computer!!! idk who is working on this project still but you guys can assign me if you want me to do it and ill change the html on the app to make sure it's clear and consistent.

• Also unrelated side note: I probably can't video conference with the client but you should ask him if he has a written description to put on the front page or any info for the about page so one of us can through it in there :)

From black_holes_frontend created by sauln : codeforgoodconf/black_holes_frontend#5

Keep track of the labels that a user gave and the labels that came from which person.

With this information, we can enable other features:

• treat given labels as votes, only confirming the label if there have been enough or unanimous votes.
• rate the accuracy of each user, with options to weight users' votes or reject them from participating.
• retract labels from users.
• enable training from #4 and track improvement.

TODO:

• Build a user model
• Build a vote model
• Connect spectra->vote with 1-many relationship
• Connect user->vote with 1-many relationship

From black_holes_backend created by amjcosta : codeforgoodconf/black_holes_backend#2

It would be ideal to create an engine with greater than 95% accuracy. One way to accomplish this may be using tensorflow to build a neural net.

Python's built-in logging module is just OK. It's a little confusing to set up and configure properly. It's still leaps and bounds better than print statements.

• Find decent logging library (maybe twiggy, logbook, or others)
• Setup logging module to keep smart log files.
• Convert all existing print statements into logging statements.
• Add more logging in all views and modules.

From black_holes_frontend created by amjcosta : codeforgoodconf/black_holes_frontend#4

We could add a "people trainer" on to the front end. The app would display a picture that has been verified as a black hole or not a black hole, and once a user guesses the answer, the app tells them whether they got it right or not.

@simian201 feel free to add more information!

From black_holes_frontend created by sauln : codeforgoodconf/black_holes_frontend#7

Tinder for Blackholes

Make the interface for voting very mobile friendly. Ideally, this would mean using swipe functionality for voting.

• Explore and demo jQuery swipe events
• Can plotly plots be swiped, or do we need to convert to a static image?
• Integrate swiping into labeller page.
• Allow users with touch screens to swipe left or right to vote yes or no.

From black_holes_backend created by amjcosta : codeforgoodconf/black_holes_backend#6

Use a minimum threshold for distinguishing between HeII and non-HeII.

From black_holes_backend created by rlbyrne : codeforgoodconf/black_holes_backend#5

Spectra that don't appear to contain the HeII spike by eye are flagged as HeII by the algorithm

From black_holes_frontend created by amjcosta : codeforgoodconf/black_holes_frontend#2

Set up a way for users to log in to their account.

From black_holes_backend created by samking314 : codeforgoodconf/black_holes_backend#3

test on
-10,000
-50,000
-100,000
-200,000
-500,000
-1,000,000

😢

After discussions with Andrew, we've decided to try converting the project to Django. We figure that in the long run, it will be easier. There will be hickups in getting everyone up to speed with the new system, but once that hurdle is crossed, expanding functionality of the frontend should be simpler.

Steps:

• Develop shell django app
• Port Galaxy model
• Port DB controllers
• Port DB seeding
• Port templates
• Port views

From black_holes_frontend created by samking314 : codeforgoodconf/black_holes_frontend#1

Write a nice readme.

• Explain the tech stack,
• Explain the features in the website,
• Provide some explanation of the neural net,
• Provide setup instructions so new users can get involved.

If you decrease the width of the browser, the nav-bar automatically collapses into a hamburger button. The hamburger button doesn't do anything though, so we can't access the navigation keys.

Fix the hamburger button is it creates a drop down navigation panel.

From black_holes_frontend created by edbreeden : codeforgoodconf/black_holes_frontend#3

Once user models are created, have the algorithm that chooses which galaxy to show a user dependent on that user.

It should

• Not show the same galaxy more than once
• Show some galaxies that we have already labeled (as a control set for the user)