Author |
Topic: Same weather conditions. |
26 replies
|
|
|
#1 posted Nov 22nd 2008, 00:31:49
|
Quote
|
I know that is not so important but :) i found that coming race in Bahrain got exactly the same weather conditions as the Indianapolis in season 9. Practice, qualify and race got every point of weather the same. There is a lot of draw possibility if the weather is draw and here we got the same. What do you think about it ??
(My english is very poor, so sory for any mistakes :] )
|
|
|
|
#2 posted Nov 22nd 2008, 00:34:17
|
Quote
|
Apparently the scripts that decide the weather aren't quite as random as they should be?
|
|
|
|
#3 posted Nov 22nd 2008, 00:34:25 (last edited Nov 22nd 2008, 00:35:02 by Dan Reed)
|
Quote
|
Unfortunately no, Its not as easy as that :p
Quote ( Toni Metsänkylä @ November 22nd 2008,00:34:17 )
Apparently the scripts that decide the weather aren't quite as random as they should be?
Depends how the randomness works :D
|
|
|
|
#4 posted Nov 22nd 2008, 00:34:50
|
Quote
|
It happens sometimes. Last season it happened too, but the final result is not always the same.
|
|
|
|
#5 posted Nov 22nd 2008, 00:37:46
|
Quote
|
I think that it will be diferent ....
|
|
|
|
#6 posted Nov 22nd 2008, 10:20:35
|
Quote
|
so what ?
Quote ( Toni Metsänkylä @ November 22nd 2008,00:34:17 )
Apparently the scripts that decide the weather aren't quite as random as they should be?
If the first weather is created randomly, it's possible that i gets created again randomly. maybe less likely, but still possible
|
|
|
|
#7 posted Jun 23rd 2020, 22:30:04
|
Quote
|
Quote ( Mikko Heikkinen @ November 22nd 2008,10:20:35 ) If the first weather is created randomly, it's possible that i gets created again randomly. maybe less likely, but still possible
Question1: What is the probability according to you to that happen? Question2: What is the probability of that happening on the same circuit? Question3: What is the probability of, once the conditions described in Question 2 occured, same scenario in race to happen?
|
|
|
|
#8 posted Jun 23rd 2020, 22:36:24
|
Quote
|
Now that's a corpse of a thread you dug up there.
1. Likely pretty high, given how many races we've had
2. A lot lower. As in, 1000^x vs 20^x
3. What?
|
|
|
|
#9 posted Jun 23rd 2020, 22:37:41
|
Quote
|
No worries, Mikko is still around to answer that :P
|
|
|
|
#10 posted Jun 23rd 2020, 22:40:07
|
Quote
|
Quote ( Constantin Heller @ June 23rd 2020,22:36:24 ) 3. What?
That not only are the forecast conditions the same as a previous race on the same track, but also those conditions play out exactly the same in the new race.
|
|
|
|
#11 posted Jun 23rd 2020, 22:52:23
|
Quote
|
Quote ( Joe Manifold @ June 23rd 2020,22:40:07 ) That not only are the forecast conditions the same as a previous race on the same track, but also those conditions play out exactly the same in the new race.
Ah. I'd place that chance close to 0. Slightly higher on tracks with less laps of course, and with less changeable conditions.
|
|
|
Quote ( Dominik Karda @ June 23rd 2020,22:37:41 ) No worries, Mikko is still around to answer that :P Yeah, and until Mikko sees this, meanwhile you can try to do it :)
|
|
|
|
#13 posted Jun 23rd 2020, 23:00:23
|
Quote
|
Quote ( Vitaly Sevov @ June 23rd 2020,22:56:51 ) Yeah, and meanwhile Mikko see this Sees it all :)
Just contemplating whether (and how) to reply
|
|
|
|
#14 posted Jun 23rd 2020, 23:01:41
|
Quote
|
Okay, to clarify, are we only talking about the rain chance here? Or are we including humidity and temperatures? Nobody will be able to give you any exact numbers because we don't know how the algorithm calculates weather for successive time slots.
|
|
|
|
|
#15 posted Jun 23rd 2020, 23:06:05
|
Quote
|
Quote ( Constantin Heller @ June 23rd 2020,23:01:41 ) Okay, to clarify, are we only talking about the rain chance here? Or are we including humidity and temperatures? Nobody will be able to give you any exact numbers because we don't know how the algorithm calculates weather for successive time slots. Everything. Including Practices, Q1, Q2 and all race weather. All the numbers, all the percentages. All the pictures of sun shinning or clouds flying in the sky...
|
|
|
For purely the rain chance, we have an easy 0.5*0.5 for the qualy weather being the same.
I'll just assume here that all rain probabilities have the same chance of occuring, because otherwise you'd have to calculate all permutations seperately, and that the range is always 5 or 10% unless you hit 0. I'll also pretend that any rain probability can follow any, which is clearly wrong but let's see.
We have 40 different probabilities each of which can be hit in the 4 race quarters, giving us 1/40^4 for all of them being the same, that's 1 in 2.56 million if you choose one race and check if another has the same weather.
Okay, we've had 1287 races, thanks /gb/ManagerProfile.asp?IDM=187
In order to see what the chance is of any two of those races having the same weather, we'll calculate the exact opposite. Picking two out of 1287 races has 827541 different possibilities. If we want none of those to have the same weather, it's (2559999/2560000)^827541 = 0.7238, so let's negate that back and we're left with a 27.62% chance of two races having the exact same weather even if it were selected completely randomly, which wouldn't make sense as weather dynamically changes.
Fun fact, if we said that the race has at least one "dry spell" out of the 4 sections with 0% rain chance, the odds already rise to 35%.
Given those numbers it's reasonable to assume that with a more realistic approach to weather calculation you'd definitely have races with the same weather from time to time.
E: Adding temperature and humidity would knock these percentages down a lot!
E2: Why am I doing this, I have an assignment due tomorrow.
|
|
|
|
#17 posted Jun 23rd 2020, 23:32:23
|
Quote
|
Because it "can" be manipulated . If you know who the script master is try pushing a few notes under the table for you're own predictions to come true :D :D :D
Though I guess it "can happen" randomly , I mean weather conditions do have more probability in certain places than others, outside of purely random conditions ?
|
|
|
Quote ( Constantin Heller @ June 23rd 2020,23:24:54 ) E: Adding temperature and humidity would knock these percentages down a lot! Exactly. So the only explanation I find is that the weather forecast is a hand made pattern. And there are x of those patterns that can be randomly assigned to.... how many circuits we have? So like that it is very similar to the so called "Birthday Paradox" Where n a set of randomly chosen people here is the set of circuits, and the birthdays are the weather forecast patterns. I bet there are more than 365...
|
|
|
Vitaly... So what's your point here?
@Constantin Heller (Ret) Lot's of numbers. Well beyond what my brain can fathom. How does that relate to the weather forecast? For those of us less inclined to put this game into quantum mathematics. heheheee :)
Hope you do well on your assignment, mate. :)
|
|
|
|
#20 posted Jun 24th 2020, 05:23:08
|
Quote
|
Quote ( Constantin Heller @ June 23rd 2020,23:24:54 ) In order to see what the chance is of any two of those races having the same weather, we'll calculate the exact opposite. Picking two out of 1287 races has 827541 different possibilities. If we want none of those to have the same weather, it's (2559999/2560000)^827541 = 0.7238, so let's negate that back and we're left with a 27.62% chance of two races having the exact same weather even if it were selected completely randomly, which wouldn't make sense as weather dynamically changes.
The first point is true. If you have 1287 events the non-repeat combinatorial picking 2 of them is 827541, and I guess you wanted to calculate how many different options are to take 2 races from the whole GPRO history without repeating.
So far so good. But then it goes on weird. The 2.56M is the combination of 40 independent events. I'm not sure how did the 40 come in order to calculate it, but there is a problem afterwards: they aren't actually independent events. We don't have a 5°-10° quarter followed by a 30°-35° quarter (and if it's season 76, good luck finding 5°-10° quarters in the first place), so the 2.56M value is faulty.
This is all for fun, tho, because in reality there is a fixed number of forecasts in the game. I don't remember how many, but let's say 100. With that number, and assuming each of those are equally likely, in 1287 races in average we'd have 13 races per weather forecast and the odds of an unique one are very small (the calculation is hard-ish, tho, because there is a nasty conditional: there can't be a repeat in a single season as far as I know).
|
|
|
|
|
#21 posted Jun 24th 2020, 10:18:18
|
Quote
|
Quote ( Edwin Silva @ June 24th 2020,05:23:08 ) I'm not sure how did the 40 come
0, 0-5, 0-10, 5-10, 5-15 and so on.
Quote ( Edwin Silva @ June 24th 2020,05:23:08 ) We don't have a 5°-10° quarter followed by a 30°-35° quarter
I'm aware of this, but it's all I could do without knowing how exactly weather is selected.
Quote ( Roy Mitchell @ June 24th 2020,04:32:47 ) How does that relate to the weather forecast?
Basically, you select one race, take its rain percentages, and then select another race and check for each number if they're the same. You do that for all possible combinations of 2 races and see if you get the same weather at least once. But as has been pointed out the math is flawed because I don't know how the weather is actually chosen.
|
|
|
I was thinking like this:
1) We had 1287 races completed. And 60 tracks in total. That makes an average of 21.45 races on same track. Let's round it to n=21. (I know there are tracks with less races completed on them)
2) Let's take as Edwin said 100 different forecasts that are created for the races. By "forecast" I understand a full pattern with all the intervals included (P,Q1,Q2,Race temp, hum, and rain) So, the probability of a given forecast occuring in a single race is p=1/100=0.01
3) Using a Binomial Distribution we compute the probability of occuring that forecast twice and get 0.01734954, and at least twice = 0.01851167. So that makes about 1.7%-1.9% chance
What do you think about that estimation?
And trying to get the answer to Question 3 that I previously proposed, suposing that appearing of the same forecast on the same track and then occuring the same scenario in the race are independent, let's say there is a weather change in same lap of a 20 laps interval. The probability of weather change on the same lap as it did before is 1/20. So we multiply the previously found probabilities by 0.05 and get 0.000867477 - 0.0009255837 which is 0.087%-0.09% These probabilities are just too low. So I think the events that I mentioned are not independent. What do you think about identical forecasts and then identical races? Coincidence?
|
|
|
|
#23 posted Jun 24th 2020, 15:18:17
|
Quote
|
These numbers are way over my head.
The only thing you've got to remember is that weather has a huge bearing on the result of the race so it can't be truly random, because that has the potential to create very unrealistic races. Imagine temperatures over 45 at the next track with very high tyre wear.
I know not everyone knew that there is a 'shopping list' of forecasts before but it seems that cat is out of the bag now. So if you are saying that we should add some new forecasts to that list then I'd agree with you.
|
|
|
Quote ( Vitaly Sevov @ June 24th 2020,14:37:29 ) p=1/100=0.01 Curiosity: S41R15 and S58R16 Anderstorp. Same forecast. 3 Wet/67 Dry and 4 Wet/66 Dry. Race time and temperature/humidity run is different. EDIT: We do not know the latter before the race. :( However, these 3 factors affect the number of D and W circles. (My research shows this.)
|
|
|
|
#25 posted Jun 24th 2020, 16:39:57
|
Quote
|
Quote ( Tibor Szuromi @ June 24th 2020,15:55:11 ) Quote ( Vitaly Sevov @ June 24th 2020,14:37:29 )
p=1/100=0.01 Curiosity: S41R15 and S58R16 Anderstorp. Same forecast. 3 Wet/67 Dry and 4 Wet/66 Dry. Race time and temperature/humidity run is different. EDIT: We do not know the latter before the race. :( However, these 3 factors affect the number of D and W circles. (My research shows this.)
Sorry, I ment that 1/100 is the probability of a given forecast to be chosen for a single race. Not to occure in it. My fault.
Regarding your example. Thanks a lot. I will check it.
|
|
|
|
#26 posted Jun 24th 2020, 17:23:53
|
Quote
|
Quote ( Vitaly Sevov @ June 24th 2020,14:37:29 ) Using a Binomial Distribution we compute the probability of occuring that forecast twice and get 0.01734954, and at least twice = 0.01851167. So that makes about 1.7%-1.9% chance
That's only the chance of one specific chosen forecast out of the 100 being selected twice, not that it happens with any one of them.
|
|
|
|
#27 posted Jun 24th 2020, 17:47:17
|
Quote
|
If I know the parachute won’t open, I probably won’t jump out of the plane.
|
|