File talk:BBFiringPointTest4.png
OK, I've got to say, I can't interpret this graph. There is just not enough information. What are the X and Y axis variables? Is the Y axis degrees (degrees of deflection from "true path")? If so, what's the X axis? Is this a histogram? If it was a histogram I would expect degrees of deflection on the X axis, maybe positive and negative with zero at the centre, and then frequency on the Y axis. But this seems to be some kind of inversion of a histogram. I am confused! Help most appreciated! Spike 16:33, 14 April 2009 (EDT)
It's not really a histogram, though it's similar to one in purpose. I can't remember what it is. It may be that a histogram would've been a better way to render the results.
The Y axis represents angles. The higher the line is at any given point, the more of an angle that point of the line represents. Note that an angle of 0 (where the shot was fired perfectly in a straight forward direction) is represented when the line hits the bottom of the Y axis.
The longer the line stays at a given height while travelling along the X axis, the more shots hit that particular angle. Keep in mind we're talking at least a thousand shots represented per graph - I wasn't sure how to label the X axis to portray this, so I didn't...
Probably didn't explain that very well on the talk page. Where the line hits the bottom of the graph, that's when shots were going straight. The area to the left of that point represents shots that went off to the left, and the area to the right represents shots that went off to the right.
You're probably thinking about now that the graph should look something more like what x^3 would look like (you can check that here if need be), because half the angles should really be "positive" and the other half should be "negative" (depending on whether they went to the left/right of the trooper). I decided to call them all positive and just bung them on opposing sides of the line.
So to make the graph, I sorted the entire list of angles in ascending order, then did a line graph of their absolute values.
Note that if you had the same chance of hitting any given angle within your maximum range, the graph would've ended up looking like a V (or a / if I'd used a negative axis).
Er, hopefully you'll be able to work out what I'm on about by graphing the test results for yourself. :)
- Bomb Bloke 22:52, 14 April 2009 (EDT)
May I venture an answer by saying that if one rotates the graph 90 degrees CCW and then flips its upper half horizontally around the point where it joins the lower half it shall become the cumulative distribution function of angular deviation, which is very amenable to statistical analysis? To demonstrate—take the V that you mentioned. After rotation it will become >, and flipping its upper half will produce /, which is the (cumulative) distribution function of the normal distribution. To avoid confusion, could you re-plot the graphs in the manner described? You will only have to use the raw (signed) angle instead of its absolute value and exchange the axes to get a sigmoid increasing from left to right. Ant 222 (talk) 00:21, 18 September 2016 (UTC)
