search

Aexyn / caliper

Automatically exported from code.google.com/p/caliper
Apache License 2.0
0 stars 0 forks source link

Correct box plots vs. 0.5-style #224

Open GoogleCodeExporter opened 9 years ago

GoogleCodeExporter commented 9 years ago 
I'll just state an opinion here and we'll see what happens.

I liked the Jesse-style box plots better. :-)

I don't really care to see outliers pinpointed. Especially since now I like to 
run many trials of many measurements, I'm being overwhelmed by black dots.

The old graphs showed me five pieces of information: namely, all five 
quartiles. That was perfect, to me!

Original issue reported on code.google.com by kevinb@google.com on 21 Feb 2013 at 5:42

GoogleCodeExporter commented 9 years ago 
I'll take it in pieces.

Outliers: we could probably hide them.  Maybe make them lighter.  Maybe just 
add an icon indicating that there _are_ outliers.  I don't think that it's a 
good idea to just make them disappear altogether.

Five pieces of information:  I think you're misremembering friendly data for a 
different graph style.  (I don't think that much of anything has actually 
changed in how it gets rendered.)  I've attached a picture of a benchmark from 
the 0.5 UI.  There certainly aren't 5 pieces of information in any of those 
bars, but that's just because the data doesn't lend itself to it.  I think that 
the new webapp is actually a huge improvement in that we actually provide 
numbers for the information that you feel like you're missing when the data 
doesn't have sufficient spread to make it show up.

I'm also trying to rework how we display detail.  I'm hoping that we'll 
actually be able to display separate graphs for things like box plots, 
histograms, linear regressions, etc. if we can use that space better.

Original comment by gak@google.com on 21 Feb 2013 at 6:43

Attachments:

GoogleCodeExporter commented 9 years ago 
I'm pretty sure your screencap just doesn't happen to include any result for 
which the quartiles are spread out enough to see.

Original comment by kevinb@google.com on 21 Feb 2013 at 8:33

GoogleCodeExporter commented 9 years ago 
Right…  I think that was my point.  Here's a screenshot of new Caliper with 
data that has spread (just one bar since that was the first one that I found.  
Unless I'm misunderstanding, we get good graphs sometimes and bad graphs other 
times in both cases.

Original comment by gak@google.com on 21 Feb 2013 at 9:03

Attachments: