Indian pitches: a veritable graveyard for bowlers!

[Guest BossBhai]

-- Removed on request of the user --

[jaideep]

the pitches dont seem to matter to the aussie quicks..!!!

[Dhondy]

Err, capitalism and marketing gimmicks too, existed before Tv networks. Duh ! Correction : Cricket is a wannabe-statistician's delight. It is a sport filled with lots of meaningless numbers and numbers that still do not get anywhere close to much meaningful conclusion from the statistics, since the field of values have far higher standard deviation than permissible in ANY sort of statistical modelling. .

Simply not true. High standard deviations simply indicate that the sample size is small, and do not invalidate the conclusion. They are used as such in many areas of science, where sample sizes are perforce small, for example for evaluating a new treatment for a rare disease. The obvious point you missed,and the one I'd have liked to see you make, is whether standard deviations should be used at all in cricket. I am sure you know that SD can only be used for datasets that have a normal distribution. You'd also have read that you cannot assume normality- if you cannot prove it, assume that your dataset has a non-parametric distribution. As far as I know, nobody has ever done a test of normality such as the Shapiro Wilks test on a cricket database. As such, talk of usefulness or otherwise of SD in such a scenario is just pseudo-science, I'm afraid.

[Dhondy]

Further, even if we were to suppose that cricket databases have normal distribution, as we are talking about means here, a more appropriate measure of dispersion would therefore be the standard error(SE), rather than SD. Standard error in fact corrects for sample size and is given by SE= SD/square root of 'n'.

[fineleg]

Wow Dhondy! A medical doctor who is also more than adept in math and statistics theory :thumbs_up: I like you :D

[Dhondy]

Please answer this question : Do you or do you not think that every statistical analysis carries an error range with it?

Nope, that's not true. Whether you do an error analysis depends on two factors- first whether you are taking a sample or looking at the whole population, and secondly whether the events you are examining are occuring randomly or not, i.e. if the event were to happen again, could the results be different? OK? Now in science, say medicine, it's impossible to look at the whole population. So you are always sampling. Secondly, when you are looking at a parameter, say, such as the mean BMI in a sample population, you are dealing with a parameter that occurs randomly, hence it is perfectly acceptable to use an error analysis in such situations. In examining cricket statistics though, error analysis are not applicable in all cases. Firstly, they are not applicable if you are, say, comparing the mean all the runs ever scored by a certain team with another- where's the sampling in this case? Secondly, not all events occur randomly in cricket and therefore error analysis should not be universally used. It would be perfectly alright to use 95% confidence intervals (CI), for example, to compare the mean of runs scored at Perth and Calcutta over a 10 year period to determine which is a higher scoring venue. In such a case, if you can prove that the runs follow a Gaussian distribution, all you have to do is add and subtract 2 SDs from the mean and present the 95% CIs. On the other hand, if you were comparing the mean of career runs scored by Lara and Tendulkar, where does error analysis come in? Firstly, it's not a sample; secondly, the runs scored are a direct reflection of the skill of the respective batsmen; where is the randomness in their genesis? How ridiculous would it look if you said, Tendulkar's career average is 55 (52-57)? Not to mention the fact that such analyses should not be used to compare batsmen who have played unequal amount of innings, as clearly the CI would be wider for those who have played less innings. The only application of an error analysis in such a scenario, and this can be used only for batsmen or bowlers with similar number of innings, is to look at their consistency (and not the adequacy or otherwise of your results). Batsmen or bowlers who are consistent, would have smaller CIs to their mean runs or wickets, again assuming normality. (For a largely skewed distribution, such as the average of Kambli's career runs, neither the mean, nor the SD would be appropriate. You'd be much better off using the median).

[Dhondy]

Wow Dhondy! A medical doctor who is also more than adept in math and statistics theory :thumbs_up: I like you :D

I am honoured, Fineleg. Alas, if only I were a decade younger...(sigh!):D

[Guest BossBhai]

-- Removed on request of the user --

[Bongosamaj]

sorry doc, you are mistaken. You will find that I've already stated my case- error analysis is a MUST for any statistics to be taken accurately, even when there is no error in sampling. Sampling error is not the same as modelling error and modelling rror + sampling error is the cumulative error in a statistical survey. I am sure you are aware of this and i'd expect our fellow 4-degree holding Bossbhai to be aware of this. Unless ofcourse, he lied about it in his usual fashion.

[Guest BossBhai]

-- Removed on request of the user --

[Bongosamaj]

First of all there is NO SAMPLING .... its the WHOLE Enchilada thats been crunched and secondly there is no modelling as there is no sample ... get that into your head if you can ... and then come back ..

Every statistical analysis is modelling, sonny. And yes, there is sampling- sampling the entire set is still sampling. These are the BASICS of statistical analysis, champ. And you claimed to have 4 degrees,including one each in Statistics and engineering ? Pffffffffft. Sorry, but you don't know the first thing about statistical analysis. And if you have the guts or the knowledge to take me on in statistics, i welcome the challenge. For i still do enjoy schooling you every now and then. In any case, there is always two sets of erros- one for the modelling ( THE THEORY YOU ARE APPLYING TO/TRYING TO PROVE- THAT is the model, champ) and one is for data-collection. Including entire set of possible numbers eliminates error in data collection- but doesn't in the theory part. Which is why, champ, even F=ma comes with error analysis, even if you use a continuous distribution integral function. Oh wait- your phoney engineer + statistics + compsci degrees won't be able to keep up in that respect. One advice- know when to quit and know when to admit your error and don't be so blatant with your lying. You are no longer 15- so try acting your age for a change.

[Guest BossBhai]

-- Removed on request of the user --

[Bongosamaj]

Nope ... its not a sample ... there is a different technical word for it in the statistical world ... lets see if you can figure it out Mr Self proclaimed statistician .

The universal set is still a sample set. First corollary of deMorgan's theorem. So much for your statistics degree, eh ? Oh and i am not a self-proclaimed statistician. I am an electronics engineer specializing in communication technology- which is all either statistics or calculus. And needless to say, my degrees arn't phony like your's.

are you confusing statistics with word play and bakwaas ... if so then you are the world leader in that ... but otherwise lets see what you got ...

Pfffft. What i got is a credible degree. Not lies and bull$hit like you who goes around claiming multiple degrees in compsci, statistics and engineering and still have time for cricket and arguing here. A person like you, who doesn't know the basic corollary of demorgan's theorems is unfit to debate statistics with me. Like i said, try to put a lid on your lies. You need far more schooling to discourse with me in statistics and most forms of mathematics. And you'd notice, Shwetabh agreed with me on the error in modelling in place and also concluded that your statistical analysis is of little value without error analysis. And guess what ? Shwetabh is not a fake-degree holder like you are.

Are you refering to the Innings defeat and follow-on thingy ?

no. The very basis of your theory ( average innings average/ average runs/test etc etc) all carry an inherent error to them,since they are in itself modelling of a phenomena ( cricket!). No analysis done on your modelling criterias = zero credibility in statistical sense. Got it, champ ? And i warn you - your googling skills and free time won't save you if you dare challenge me in statistics or engineering. For its very easy to see that you are a phony when you claimed degrees in that field.

and I told you that I can provide stats for those scenarios but not if you are going to act like spoilt kid and still go around proclaiming that stats = bakwaas and mean nothing in sports.

A statistical model without error analysis is bakwaas, will always be bakwaas and will always get you a big fat ZERO in a stats class. Comprende, fake-degree-man ? What is even laughable is that you have no idea to go about doing error analysis for your lil stats project. A fake-wannabe number-cruncher like you clearly has no idea about how to proceed with statistical analysis. Anyways, office is closing, its time for me to go home and enjoy my weekend. Little time for me to bother with fake degree holders and liars like you.

[Bongosamaj]

Kya hua boss ? still feel like challenging me in statistics or do you concede ? Busy looking for a ghostwriter statistician to bounce your pending statistics grilling by me ? Brushing up on deMorgan's theorem ? Bossbhaga....perhaps that should be your new name.

[Guest BossBhai]

-- Removed on request of the user --

[Bongosamaj]

the only DeMorgans theorem that I know is to do with boolean logic ... not (x AND y) = (not x ) OR (not y) so its corollary would be not (x OR y) = (not x ) AND (not y) Now enlighten me on how this applies in the topic we are discussing

if you need enlightenment on how deMorgan's theorem relates to statisical analysis, i am sorry to say, you have not taken even ONE course in stats, nevermind your ridiculous claims of having a stats degree. That is the basis of probability analysis and stochastic modelling, champ, since it defines precisely what a set is, what a set is not and ALL statistical sampling methods rely fundamentally on set theory to define their sample space and set relations (deMorgan's theorem) defines what is/is not admissable data for your stochastic modelling. Oh btw, deMorgan's theorem YOU quoted is a digital logic adapted version of SET THEORY. Which is originally where deMorgan stated his theorem and is applied to statistics as all statistical analysis is derived fundamentally from set-theory. next time, think twice before trying to bull$hit me in your qualifications. Comprende ?

[Guest BossBhai]

-- Removed on request of the user --

[Bongosamaj]

Me : Iam presenting the Entire Dataset its not a sample drawn from somewhere you : there is sampling- sampling the entire set is still sampling Me : No its not a sample its the whole thing. You : The universal set is still a sample set. First corollary of deMorgan's theorem me : heres the Demorgan thoerem . I dont see what boolean logic has got to do with sampling.

Very simple, champ. DeMorgan's theorem is made in SET THEORY. Corollary of deMorgan's theorem is that the universal set is a set by itself. Sampling the entire universal set is still sampling, where sample space A = U, where U is the universal set and A is all the subsets of universal set. A set is a subset of itself- that is, same as a number is always divisible by itself. If this much is not clear to you, i am sorry to say, you know ZERO about statistics. Time for you to actually attend a class or two in it, instead of bragging that you got degrees in something you are hopelessly inept at.

Ignoring the usual rants for a minute .

hahahaha..i am sure you'd love to ignore the FACT that you claimed multiple degrees in computer science, engineering and statistics.

[Guest BossBhai]

-- Removed on request of the user --

[Bongosamaj]

that is not demorgans theorem that is a set theory ... by Venn (I forget his first name) IIRC .. Demorgans work was completely different .. and I have shown it in my previous post.

DeMorgan's theory IS to do with set theory! DeMorgan did NOT write boolean logic theorem- that is boolean logic's ADAPTATION of deMorgan's theorem! George boole made his theory of Boolean algebra 25 yrs after deMorgan stated his theorem, so how can deMorgan's theorem have to do with Boolean logic, considering it was before boolean logic even EXISTED ? FYI, deMorgan defines what a universal set is and what subsets of the universal sets are- IN HIS THEOREM. This theorem was used by John Venn more than 25 years later to formulate his VENN diagrams ( he is NOT the father of set theory, deMorgan is, as deMorgan DEFINES what a set is and interrelation of sets in the first place!). nice try champ- but still, no cigar.

lets get to the bottom of this ... heres the Demorgan theorem not (x OR y) = (not x ) AND (not y) you said this was applicable to what we were discussing so what is x and y in our example and why are we doing a OR operation on them and doing a not ...

What is X and Y ? If you answer that and you know your set theory, you'd know the connection. But googling will take you only so far.