LFS stats!

[smith]

Hello. Some of you already used my insim stats 1.0 tool to create a bit of statistical information about races in S1 and S2 alpha demo times.
With full S2 alpha release I did some modifications/bugfixes to it, so updated version is rolled out for your judgement.
Detailed usage instructions are in readme, so please check it first.

Example of output: http://www.kursknet.ru/~smith/stats/results_qual.html and http://www.kursknet.ru/~smith/stats/results_race.html .

graph v1.19
- Fixed zero-lapped player case;
- Graphs are not created if they already exist on disk and have non-zero length;
- (Hopefully) fixed sometimes wrongly placed laptime chart title;

stats v1.49
- User name replaced player name in tsv macro;

[Phlos]

It's just awesome. Thank you for your work

[Shorty]

very useful this script. Great work!

[sebastiano]

thx

[snewham]

Originally Posted by sebastiano

thx

ditto

[WiRuZik]

I have a problem with lfs stats.. i have many replays but it doesn't work on each of them.. for example our last league race.. long race is ok (results are a little wild) but it doesn't make stats from short race (5 laps).. when the replay start, it shows Waiting for connection and at the end of the race the stats aren't made..

[Palee]

we also like and use your product.

if i may to suggest one thing:

Average lap and consistency feature could possibly disregard laptimes of pits and first lap?

Or a graph where the lowest value always the last positioned racer, and the highest value is the race leader?
You know, a 0% to 100% graph which would better demonstrate the relative distance of racers, from start to finish.

[Jeddix]

Thanks

[sebastiano]

I have a small problem with the graph, when I execute it I give the following error: Error Cannot create graph/11,08,05_142309_lbl.png.I have installed the microsft runtime, I have executed the graph in the directory "stats" but I have always the problem. How I can resolve it?

thx

[ekze]

Originally Posted by sebastiano

I have a small problem with the graph, when I execute it I give the following error: Error Cannot create graph/11,08,05_142309_lbl.png.I have installed the microsft runtime, I have executed the graph in the directory "stats" but I have always the problem. How I can resolve it?

thx

create "/graph" directory

[sebastiano]

thx

[smith]

Originally Posted by WiRuZik

it doesn't make stats from short race (5 laps).. when the replay start, it shows Waiting for connection and at the end of the race the stats aren't made..

I need more info, what messages you see, and where are they appear.

Originally Posted by Palee

Average lap and consistency feature could possibly disregard laptimes of pits and first lap?

I'm skipping first lap already. Can't skip pit laps, as there are no pit-related insim packets (yet).

Originally Posted by Palee

Or a graph where the lowest value always the last positioned racer, and the highest value is the race leader?

I do not undersntand what you mean, can you draw some example?

[RudiTurbo]

smith, could the average lap thingy just disregard laps that go bout 25 secs over the average, that would count out the pit laps, would also count out the mistakes, but it would nicely show who had the fastest pace if he had done a clean race
Because average laps and average speeds are in most cases always the same as the results anyway, we dont need to see the results 2 or 3 times, the results table is enough, that average thingy which would count out the laps that go 25 secs over average would help us to do some more speculations
or the time could be 10 or 15 secs even, cause the stops for formulas are shorter.

[ekze]

Originally Posted by RudiTurbo

smith, could the average lap thingy just disregard laps that go bout 25 secs over the average, that would count out the pit laps, would also count out the mistakes, but it would nicely show who had the fastest pace if he had done a clean race
Because average laps and average speeds are in most cases always the same as the results anyway, we dont need to see the results 2 or 3 times, the results table is enough, that average thingy which would count out the laps that go 25 secs over average would help us to do some more speculations
or the time could be 10 or 15 secs even, cause the stops for formulas are shorter.

This must be great.

[Palee]

Originally Posted by smith

can you draw some example?

sure.

here is what you already have in place:
http://teaminferno.hu/p/race1003Acsop.png
and here is the normalized (relative distances between the drivers)
http://teaminferno.hu/p/race1003Acsoprelativ.png
so the first part of the race is quite visible

[Gimpster]

Smith,

Very cool app. I am in the process of trying to put togeather a new race series for the LFS comunity. One thing I have identified that I would like is a tool to track qualifying across several different servers over an extended period of time and to track the race progress and results as well. Idealy I would then like to display this data in real time to a website. I have no experance in programing or website building and juct wondering if this could be done? If it can some pointers or a susggestion of who might have the time, energy and willingness to do it would be great.

[smith]

Originally Posted by RudiTurbo

smith, could the average lap thingy just disregard laps that go bout 25 secs over the average, that would count out the pit laps, would also count out the mistakes, but it would nicely show who had the fastest pace if he had done a clean race

I am going to add pit-stop laps skipping when insim'll be able to send pit packets.
Skipping laps slower than average by fixed amount of time is not very good idea imho.

Originally Posted by Palee

here is what you already have in place:
http://teaminferno.hu/p/race1003Acsop.png
and here is the normalized (relative distances between the drivers)
http://teaminferno.hu/p/race1003Acsoprelativ.png
so the first part of the race is quite visible

Looks like race1003Acsoprelativ.png main idea is exactly the same as my lap by lap graph, with the only difference that it's bound to time, and my graph is bound to positions.

Originally Posted by Gimpster

One thing I have identified that I would like is a tool to track qualifying across several different servers over an extended period of time and to track the race progress and results as well. Idealy I would then like to display this data in real time to a website. I have no experance in programing or website building and juct wondering if this could be done?

Everything is possible. Take any insim lib (for example, sdether or Messiah one) and you're almost there, you just need to spend some time. Also, remember the main principle: if you want anything to be done, do it by yourself

[RudiTurbo]

Originally Posted by smith

I am going to add pit-stop laps skipping when insim'll be able to send pit packets.
Skipping laps slower than average by fixed amount of time is not very good idea imho.

Can You bring Me a reason?

[Hoellsen]

Is it in any way possible to make the stats relay compatible? That way you could run different insim applications like LFS Spectator AND the stats at one time.

[Da Hoe]

because it's about the average lap, rudi. and what is an average lap when you dont count the laps in which you made faults?. imho every lap should be counted.

[HS-78]

Nice tool that one - could you give a additional diff row in the table of the best possible lap? I mean difference to your own real best lap in that race.

[RudiTurbo]

Originally Posted by Da Hoe

because it's about the average lap, rudi. and what is an average lap when you dont count the laps in which you made faults?. imho every lap should be counted.

imho average speed and laptime stats have always been pointless, cause they give exact same positions as the finishing order. The best stats are the ones which give You comparison from different sides of view and speculation.

[Da Hoe]

ok, but then you'd need another name for it :P

and i still like the average thingies in the stats though.

[inCogNito]

Originally Posted by RudiTurbo

imho average speed and laptime stats have always been pointless, cause they give exact same positions as the finishing order.

btw that isn't right. unless smith changed it in the latest version

[smith]

Originally Posted by RudiTurbo

Can You bring Me a reason?

Tracks have different pit-lane length, different pitlane entry/exit length and speed. Finally, pit-stop time length can vary very significally too...
And, what's more important, I can't distinguish whether one heavily spun or pitted in...
The best solution is to wait for pit-stops packets.

Originally Posted by Hoellsen

Is it in any way possible to make the stats relay compatible?

Afaik there's still no relay for S2 yet. Thus, I can't test that.

Originally Posted by HS-78

could you give a additional diff row in the table of the best possible lap?

Sure, that'll be done.

Originally Posted by inCogNito

btw that isn't right. unless smith changed it in the latest version

Thats right, avg. lap order reflects exact finish positions for finished ones.

[RudiTurbo]

Originally Posted by inCogNito

btw that isn't right. unless smith changed it in the latest version

It can be diff, if someone starts from last and finishes 3rd, but that's what we have the position graph for.

[inCogNito]

Originally Posted by smith

Thats right, avg. lap order reflects exact finish positions for finished ones.

Well, i showed you a result where the finishing order was different than the avg. lap order, and you said it's OK, because how the avg. lap is calculated. (not just race length / laps)

maybe a misunderstanding back then

[smith]

Originally Posted by inCogNito

Well, i showed you a result where the finishing order was different than the avg. lap order, and you said it's OK, because how the avg. lap is calculated. (not just race length / laps)

It was changed since that conversation, as I realised I did math in the wrong way.

[D34N0]

Smith - Are you able to determine if a valid pit stop has taken place or the person has gone straight to the garage and out (I know doing this resets the laps to 0) as it would be nice if Stats was able to detect this then placing the person as a DNF at the bottom with 0 laps completed.

The reason I am asking is during a league race someone sustained damage then immediately went to the garage came out and completed the race. And was fewer laps down than others who had made legitamate pit stops. This was found out after watching the replay and amending the stats page

I have also added a post in the improvement section asking for the Shift+S and Pits options be allowed to be switched off so there is no going to the garage during a race just pit stopping.

Regards
Dean

[bobvanvliet]

I have an improvement suggestion: i find the "lap by lap positions" graph to be somewhat unclear, because every line's plot positions are based on the progress of THAT racer. In stead, I think it would be better to plot the positions against the race progress of the leading car, so that events that are at the same lap-axis position, actually happened at the same time. Now when people are down on or more laps, their line is no longer synchronous with the race leaders' lines.

You catch my drift?

Anyhoo, great piece of work!