Anonymous user
Talk:Averages/Mean angle: Difference between revisions
→Accuracy: added comments about amplitute of zero, any angle... and the use of exact values instead of near values for ATAN2. -- ~~~~
(→Accuracy: It has its uses.) |
(→Accuracy: added comments about amplitute of zero, any angle... and the use of exact values instead of near values for ATAN2. -- ~~~~) |
||
Line 20:
:::Take a look at [[wp:Directional statistics]] and the other pages it might link to, as well as talk pages. I got the impression that this is not for all angular measurements - sometimes the number of turns is significant; and it is easy to get answers that are computed correctly but without checking the amplitude of the resultant mean of vectors, you miss out on information on what credence to give to the result. What is the mean of 0 and 180 degrees? the amplitude of the resultant is zero so any angle returned is of no import.
:::There does seem to be situations where the formula is applied to give meaningful results. You might be stimulating a species of land mammal in a controlled environment and monitoring in which direction they, ran for example. --[[User:Paddy3118|Paddy3118]] 06:15, 12 July 2012 (UTC)
:::: (Regarding an amplitude of 0, angle returned is of no import) --- yes, I noticed that when the REXX trig functions used short-circuiting and returned exact values instead of near values, this made a significant difference on how '''ATAN2''' ''interpreted'' the results, as it uses the sign of the arguments to figure out what quadrant it's in, and '''-8.154E-61''' (for ATAN2) is much different than zero, albeit a very small difference from zero. -- [[User:Gerard Schildberger|Gerard Schildberger]] 20:47, 23 July 2012 (UTC)
| |||