Variable View about dug HOT 12 CLOSED

• clicking variable in study drop down - launch new tab in background automatically?
• Variables according to DUG Score - change to Variables according to Dug Score
• also what is Dug score?
• when i highlighted Variables according to DUG Score it highlighted all the text in the popup window that hovers over the variables and i was unable to change cursor from the + symbol and interact with the webpage at all; had to delete URL back to just https://helx-ghench.apps.renci.org/ and then it crashed
• studies seem like they should be listed by # of variables - are they instead listed by their 'Dug score'? can we explain this somewhere?

from dug.

Things I'm trying to reproduce:

• when i highlighted Variables according to DUG Score it highlighted all the text in the popup window...then it crashed << I'm not able to reproduce this. I was able to highlight the text, "Variables according to DUG Score" without any issues. I'm guessing the specifics of this problem are superficial to the same time lag effect I've experienced a few times when I've been doing lots of zooms & highlights in quick succession.

from dug.

Questions for Hannah:

• clicking variable in study drop down - launch new tab in background automatically?

I do think they should open in a new tab and I can implement that change quickly. However, I'm not sure about "in background automatically" - maybe this is something Matt's done for other apps?

• also what is Dug score?

This is def something that the DUG search team needs to define for a user audience Programmatically, the DUG score is based on the Elastic Search Score with some extra weighting that is implemented in the variable search API endpoint function encoded in the DUG repo. Kira suggests we talk to Chris Bizon about this too.

• studies seem like they should be listed by # of variables - are they instead listed by their 'Dug score'? can we explain this somewhere?

Yes, the study order is based on the score of the variables they hold... This is because we're actually scoring the variables and then clustering them based on their study
Yes, we can have some explanatory text somewhere - what is your preference? Again, this is probably a good things to coordinate with Matt

from dug.

Kira's comments

• threshold
• new tabs for each zoom view
• check boxes instead of push pins
• check boxes for the variables

from dug.

• Fix the variable count that shows in top line of both views

from dug.

Some ideas of how the view changes as a user interacts with it:
User goal: I want to combine 2+ datasets together, but I want to do so with the least amount of work possible.
User flow: type in a concept that they are investigating (ex: heart attack, asthma, opioid-induced hyperalgesia) > search > select toggle view to see which variables scored the "highest" on the Dug score > zoom into the top ~10% of scores > opens a new tab with a histogram view of variables in the top 10% > I go through the table and select either specific variables (checkbox) or studies (checkbox) > have an option that allows me to compare data dictionary similarity (NEW FUNCTIONALITY) or allows me to figure out how easy it is to combine/harmonize these datasets (NEW FUNCTIONALITY) > OR I choose to keep these two studies selected and now I want to view ALL the variables associated with those two studies that are associated with the concept that I originally search for > new window of histogram viewer so I can see all the variables and their dug score (what are all the related variables that I should be looking at?)

from dug.

• functionality where user can click on bar that has variable that they are interested in

from dug.

need to be able to define Variables according to DUG Score on page for users

from dug.

Extra histogram options would be awesome to have in the future. For reference, something like matplotlib allows the user to reset zoom, go back/forward in your zooms, save the figure, and other adjustments.

Also, I just happened to stumble upon this, but there's histRef.current.getChart().downloadImage(name: str) so this would probably be pretty straightforward to add.

from dug.

@hhiles, Griffin and I are reviewing the notes on this issue ticket.

• Do you remember what you meant by "search within search" functionality?
• Do you have a definition for DUG Score - if not, would Guarav or Yaphet? Here's a link to the boosting on the API endpoint, search.py#L307
• Some kind of popup/help text to explain functionality of the push pins >>> "Select to highlight variables from this study"

from dug.

@gingin77

1. "search within search" was the ability for users to narrow their search. Eg, I could run a search for diabetes and then search within those results again either with a second search, with adding filters, removing results, etc... some kind of functionality for filtering.
2. John Cheadle was the one we were depending on for that in the past; @gaurav or @YaphetKG would be my second guess.
3. I like that description for the push pins!

from dug.

Hi @gingin77 , i believe the dug scoring currently relies on Elastic Search scoring. Here is an article and from elastic docs explaining how that works. In essence its quantifying how relevant a document is based on the query term, by weighing (boosting) attributes of a dug element based grading we have assigned to them (John Cheadle and Alex originally determined this boosting values).

from dug.