Evan Huus
changed
bug 8703
| What |
Removed |
Added |
| CC |
|
[email protected]
|
| Hardware |
x86-64
|
All
|
| Version |
1.8.7
|
SVN
|
| OS |
Windows 7
|
All
|
| Severity |
Normal
|
Minor
|
Comment # 1
on bug 8703
from Evan Huus
For the first case we could potentially do something special when we don't have
a statistically significant number of packets - I'm not sure what that
'something special' would be though. Maybe just put the rate as N/A and then
put an explanation in a tooltip?
The second case is much trickier - there are statistical techniques for
removing outliers, but they're complicated, not always appropriate, and
probably wouldn't work in your case anyways because there are so few packets.
The simplest thing that might work for the second case would be to, instead of
reaching a total time by summing the available times, take their median and
multiply it by the sample size. This would produce a rate of ~1400 b/sec by my
calculation.
However, we probably don't want to do that by default; it would have to be a
preference, at which point it might just be simpler for people to export the
values as CSV and do the calculations they want in Excel or a similar product.
You are receiving this mail because:
- You are watching all bug changes.