711990 - [Telemetry front-end] Show bucket heights as percentages, not raw values

Status: RESOLVED FIXED

(Mozilla Metrics :: Frontend Reports, defect)

Description

[Justin Lebar (not reading bugmail)]

In general, we're much more interested in the percentage of the total that a bucket represents than in the raw number of samples in a bucket.

In fact, the only time I can imagine I'd be interested in the raw number of samples is when the histogram has very few samples; in this case, giving me the raw number would serve as a warning that this data is unreliable.

Of course, the number of samples is displayed at the top of the page, but this is pretty far away from the graph itself; in fact, it's above the graph's title.  If the number of samples were displayed closer to the graph (bug 711989), then I can't think of any scenario where it would be necessary to display raw numbers of samples in the histogram itself.

Instead of raw counts, the histogram's y-axis and rollovers give percentages.  Two significant figures should be more than enough here.

Comment 1

[Paulo Pires]

Take a look if it is ok now.
When comparing, each serie has his maximum used to normalize the data separately.

Comment 2

[Justin Lebar (not reading bugmail)]

> In general, we're much more interested in the percentage **of the total**

I was thinking that we'd normalize it so that all the buckets add up to 100%.

I might leave the Y-axis labels as raw values, since there's no harm in that.  (Although "7.5m" is preferable to "7,500,000".)

Comment 3

[Paulo Pires]

Normalization is fixed.

Comment 4

[Justin Lebar (not reading bugmail)]

I like it this way, although I can understand if others don't and want you to change it back.  If you changed it back, I wouldn't miss it if we had a CDF.  :)

Comment 5

[Annie Elliott]

updating whiteboard

Comment 6

[Annie Elliott]

updating status and assignee

Comment 7

[Annie Elliott]

New JIRA migration identifier - 55 BZ issues being considered in current planning for Q3