SandboxFlattener/Graphs

These graphs are drawn with iiley's SmogPainter application from data gathered by iiley's StatistRobot bot. The graphs are segmented on distances. I have run StatistRobot against MakoHT (since that was what Tango asked for). And, for reference, SandboxDT (the king of flattening) and Mako since it hosts the exact same guns as MakoHT and it illustrates why MakoHT places 10+ positions ahead of Mako in EternalRumble. Actually the difference in movement gives an even bigger advantage than that. Mako1.2.4+ saves enemy data between matches, which gives it an extra edge in EternalRumble.-- PEZ

Please don't ask me for more profiling graphs. Instead encourage iiley to upload StatistRobot and SmogPainter. -- PEZ

---

Distances 91-250

Distances 251-490

Distances 491-810

---

It's the read curve I look at. I have never figured what the blue curve is about. =) -- PEZ

I've always thought the blue curve was the same as the red curve but with a larger factor of change each VB hit. Or it could be just the latest 50 VBs or something. -- Tango

>I've always thought the blue curve was the same as the red curve but with a larger factor of change each VB hit.

Yes,you are right. -- iiley

Yipee!!!!!!!! -- Tango

I still don't get it. Sorry for my thick head. "Larger factor of change each VB hit" ¿que? -- PEZ

Basically, everytime a VB hits it adds a certain amount to the probitility score thing for the bearing it was fired at. For the blue line that certain amount is higher. Does that make any more sense? -- Tango

Shouldn't this mean that the values for the blue one are n*values for the red one? I can't seem to find an n for wich this holds ...

My explanation was a bit of a simplification. There's probably some smoothing process on the red one, and the blue one probably uses only the latest few VB hits, hence it having lots of high peaks and lots of low troughs, because some of the bearings won't have had any hits, and some won't have had any misses. -- Tango (PS NB: i've never done anything like this, so i'm only guessing, but iiley said i was right, so...)

You are all wrong,hehe~~~;} and i said you Tango was right not mean your "Or it could be just the latest 50 VBs or something",is your "I've always thought the blue curve was the same as the red curve but with a larger factor of change each VB hit. "

I dont know how to explain it in English,i just said these: blue and red curves value is a array in double (blue=new double[101])

blue[index]: index is the angle index,example index=50 means the shoot bearing is 0 degrees,0 means the shoot bearing is -50 degrees,and 100 means 50 degrees. blue[index]'s value is 0 when initialized,then when the VB start hit,if a bullet with 0 degrees shoot angle hited,blue[50]+=0.01,and others*=0.99.(this is same to DuelistMicro's,you can have a look into DuelistMicro,then you can know what blue[] means).

red[index]:index is same to blue's. red[index]'s value is 0 too when initialized,then when the VB statt hit,if a bullet with 0 deredds shoot angle hited,red[50]+=0.01,and others dont change.so it is flater than blue curve.;]

I will release StatistRobot and SmogPainter's source code soon,then you can understand how it works clearly.Hi,PEZ,can you copy the source which i sended to you to robowiki then give a link then we can download it,i am not in home so i cannot get the source this time,Thanx. -- iiley

OK. Do you want me to the StatistRobot and SmogPainter you sent me downloadable? Just say "yes" and I'll do it right away. -- PEZ

and...dont compare red curve and blue curve,they are all div a number before drawing,but not dived the same number.so...only same color curve can be comparable. -- iiley

Let's see if I understand this: The blue one is a rolling average while the red one is just an accumulation ...

Yes,and ...Yes.;] -- iiley