Variations in GN'R's setlists

[SoulMonster]

Unfortunately, although any effort is laudable, this doesn't calculate anything of value

I am not going to say it is "valuable" per se, but I do find it interesting to compute the similarity scores for tours and years and it seems like others agreed. A bit arrogant of you to claim this is of no value to us, no? Do you disagree that the methodology used gives an accurate number of the amount of difference between two set lists and hence can be used to demonstrate the amount of variation during specific tours and years?

[bbbba]

Unfortunately, although any effort is laudable, this doesn't calculate anything of value

I am not going to say it is "valuable" per se, but I do find it interesting to compute the similarity scores for tours and years and it seems like others agreed. A bit arrogant of you to claim this is of no value to us, no? Do you disagree that the methodology used gives an accurate number of the amount of difference between two set lists and hence can be used to demonstrate the amount of variation during specific tours and years?

Apologies for sounding condascending, but we all stand on the shoulders of giants. If you are keen on solving these kinds of problems, this is the holy grail

Papoulis (2002) Probability, Random Variables and Stochastic Process

[SoulMonster]

Unfortunately, although any effort is laudable, this doesn't calculate anything of value

I am not going to say it is "valuable" per se, but I do find it interesting to compute the similarity scores for tours and years and it seems like others agreed. A bit arrogant of you to claim this is of no value to us, no? Do you disagree that the methodology used gives an accurate number of the amount of difference between two set lists and hence can be used to demonstrate the amount of variation during specific tours and years?

Apologies for sounding condascending, but we all stand on the shoulders of giants. If you are keen on solving these kinds of problems, this is the holy grail

Papoulis (2002) Probability, Random Variables and Stochastic Process

I don't think I will bother to spend time learning statistics again, why don't you rather point out why the methodology lined out in the first post isn't a good way to compare the similarities between individual setlists?

EDIT: No, wait. Not why it isn't "a good way", but why it is doesn't give "anything of value".

Edited October 12, 2012 by SoulMonster

[bbbba]

Unfortunately, although any effort is laudable, this doesn't calculate anything of value

I am not going to say it is "valuable" per se, but I do find it interesting to compute the similarity scores for tours and years and it seems like others agreed. A bit arrogant of you to claim this is of no value to us, no? Do you disagree that the methodology used gives an accurate number of the amount of difference between two set lists and hence can be used to demonstrate the amount of variation during specific tours and years?

Apologies for sounding condascending, but we all stand on the shoulders of giants. If you are keen on solving these kinds of problems, this is the holy grail

Papoulis (2002) Probability, Random Variables and Stochastic Process

I don't think I will bother to spend time learning statistics again, why don't you rather point out why the methodology lined out in the first post isn't a good way to compare the similarities between individual setlists?

EDIT: No, wait. Not why it isn't "a good way", but why it is doesn't give "anything of value".

The results are of no value since you don't use the correct approach. This is an advanced topic in data processing and the text I referred to is mandatory reading for graduates in scientific and engineering disciplines. It describes the correct method.

[SoulMonster]

Unfortunately, although any effort is laudable, this doesn't calculate anything of value

I am not going to say it is "valuable" per se, but I do find it interesting to compute the similarity scores for tours and years and it seems like others agreed. A bit arrogant of you to claim this is of no value to us, no? Do you disagree that the methodology used gives an accurate number of the amount of difference between two set lists and hence can be used to demonstrate the amount of variation during specific tours and years?

Apologies for sounding condascending, but we all stand on the shoulders of giants. If you are keen on solving these kinds of problems, this is the holy grail

Papoulis (2002) Probability, Random Variables and Stochastic Process

I don't think I will bother to spend time learning statistics again, why don't you rather point out why the methodology lined out in the first post isn't a good way to compare the similarities between individual setlists?

EDIT: No, wait. Not why it isn't "a good way", but why it is doesn't give "anything of value".

The results are of no value since you don't use the correct approach. This is an advanced topic in data processing and the text I referred to is mandatory reading for graduates in scientific and engineering disciplines. It describes the correct method.

Hehe, I have never claimed there aren't other methods more suitable for calculating similarities between lists (I was limited by those formulas available to me in Excel, where all my setlists are stored), but if you believe there are ONE correct approach to measuring similarities, then you have a lot more to learn. My question to you remains: Why isn't the methodology I have used suitable for quantifying setlist similarities? If you keep on avoiding this question but rather insist on referencing to textbooks used for undergraduate courses in probability theory, then I have no option but to conclude you have no idea what you are talking about. Don't misunderstand me, I believe you probably have taken a course or two in statistic, I just don't think you really understand what I have calculated nor how I have done it.

Edited October 12, 2012 by SoulMonster

[bionic56]

IS !!!! this a book worm forum or a gnr forum ?????????

Thankyou Sir !

[bbbba]

Unfortunately, although any effort is laudable, this doesn't calculate anything of value

I am not going to say it is "valuable" per se, but I do find it interesting to compute the similarity scores for tours and years and it seems like others agreed. A bit arrogant of you to claim this is of no value to us, no? Do you disagree that the methodology used gives an accurate number of the amount of difference between two set lists and hence can be used to demonstrate the amount of variation during specific tours and years?

Apologies for sounding condascending, but we all stand on the shoulders of giants. If you are keen on solving these kinds of problems, this is the holy grail

Papoulis (2002) Probability, Random Variables and Stochastic Process

I don't think I will bother to spend time learning statistics again, why don't you rather point out why the methodology lined out in the first post isn't a good way to compare the similarities between individual setlists?

EDIT: No, wait. Not why it isn't "a good way", but why it is doesn't give "anything of value".

The results are of no value since you don't use the correct approach. This is an advanced topic in data processing and the text I referred to is mandatory reading for graduates in scientific and engineering disciplines. It describes the correct method.

Hehe, I have never claimed there aren't other methods more suitable for calculating homology between unordered lists, my question to you remains: Why isn't the methodology I have used suitable for quantifying setlist similarities? If you keep on avoiding this question but rather insist on referencing to textbooks used for undergraduate courses in probability theory, then I have no option but to conclude you have no idea what you are talking about. Don't misunderstand me, I believe you probably have taken a course or two in statistic, I just don't think you really understand what I have calculated nor how I have done it.

This is a probabilistic problem - if you don't see this as a probabilistic problem then you can't process any data. The data is random and subject to probability - the probability distribution of track listings is equal to the normalized histogram. From this, you can determine error limits, standard deviations, confidence levels etc which is the conventional way of analyzing data.

[SoulMonster]

Unfortunately, although any effort is laudable, this doesn't calculate anything of value

I am not going to say it is "valuable" per se, but I do find it interesting to compute the similarity scores for tours and years and it seems like others agreed. A bit arrogant of you to claim this is of no value to us, no? Do you disagree that the methodology used gives an accurate number of the amount of difference between two set lists and hence can be used to demonstrate the amount of variation during specific tours and years?

Apologies for sounding condascending, but we all stand on the shoulders of giants. If you are keen on solving these kinds of problems, this is the holy grail

Papoulis (2002) Probability, Random Variables and Stochastic Process

I don't think I will bother to spend time learning statistics again, why don't you rather point out why the methodology lined out in the first post isn't a good way to compare the similarities between individual setlists?

EDIT: No, wait. Not why it isn't "a good way", but why it is doesn't give "anything of value".

The results are of no value since you don't use the correct approach. This is an advanced topic in data processing and the text I referred to is mandatory reading for graduates in scientific and engineering disciplines. It describes the correct method.

Hehe, I have never claimed there aren't other methods more suitable for calculating homology between unordered lists, my question to you remains: Why isn't the methodology I have used suitable for quantifying setlist similarities? If you keep on avoiding this question but rather insist on referencing to textbooks used for undergraduate courses in probability theory, then I have no option but to conclude you have no idea what you are talking about. Don't misunderstand me, I believe you probably have taken a course or two in statistic, I just don't think you really understand what I have calculated nor how I have done it.

This is a probabilistic problem

No, this is a similarity problem. I am not computing the probabilities of any given setlist happening, I am computing the similarities between two setlists that have happened.

But again: Explain to me why the outlined methodology doesn't give us scores that directly reflect the amount of similarity between two setlists. If you really can explain why the similarity scores I have obtained doesn't correlate to the differences in the setlists, then I have no option but to admit that the methodology is flawed. It really is that simple. It is a test to find out if you really know what you are talking about or if you just have some knowledge of probability theory.

[bbbba]

Unfortunately, although any effort is laudable, this doesn't calculate anything of value

I am not going to say it is "valuable" per se, but I do find it interesting to compute the similarity scores for tours and years and it seems like others agreed. A bit arrogant of you to claim this is of no value to us, no? Do you disagree that the methodology used gives an accurate number of the amount of difference between two set lists and hence can be used to demonstrate the amount of variation during specific tours and years?

Apologies for sounding condascending, but we all stand on the shoulders of giants. If you are keen on solving these kinds of problems, this is the holy grail

Papoulis (2002) Probability, Random Variables and Stochastic Process

I don't think I will bother to spend time learning statistics again, why don't you rather point out why the methodology lined out in the first post isn't a good way to compare the similarities between individual setlists?

EDIT: No, wait. Not why it isn't "a good way", but why it is doesn't give "anything of value".

The results are of no value since you don't use the correct approach. This is an advanced topic in data processing and the text I referred to is mandatory reading for graduates in scientific and engineering disciplines. It describes the correct method.

Hehe, I have never claimed there aren't other methods more suitable for calculating homology between unordered lists, my question to you remains: Why isn't the methodology I have used suitable for quantifying setlist similarities? If you keep on avoiding this question but rather insist on referencing to textbooks used for undergraduate courses in probability theory, then I have no option but to conclude you have no idea what you are talking about. Don't misunderstand me, I believe you probably have taken a course or two in statistic, I just don't think you really understand what I have calculated nor how I have done it.

This is a probabilistic problem

No, this is a similarity problem. I am not computing the probabilities of any given setlist happening, I am computing the similarities between two setlists that have happened.

But again: Explain to me why the outlined methodology doesn't give us scores that directly reflect the amount of similarity between two setlists. If you really can explain why the similarity scores I have obtained doesn't correlate to the differences in the setlists, then I have no option but to admit that the methodology is flawed. It really is that simple. It is a test to find out if you really know what you are talking about or if you just have some knowledge of probability theory.

If you don't know why this is wrong, please go to a mathematician and explain the problem to them. Listen carefully to their approach to the problem and the correct interpretation of data. You will hear them mention the words random variable and probability distribution as it is explained to you. If these are alien terms to you, then you are not the one to solve this problem.

[SoulMonster]

If you don't know why this is wrong, please go to a mathematician and explain the problem to them. Listen carefully to their approach to the problem and the correct interpretation of data.

Uhm, YOU are the one saying the methodology is wrong, you should be able to explain why it is wrong . So again: Why won't the methodology outlined in the first post succeed at giving numerical values for the amount of similarity between two setlists? The methodology is remarkably simple, it should be very easy to point out what is wrong with it rather than just say it is without value and that another method is better Come on now, stop wasting my time!

EDIT: And since it seems to have passed you by: I do know a little about statistics, including probability theory. Actually, I have an academic major in mathematics with about half of the courses being statistics. I admit this is more than 10 years ago but I do like to think I remember quite a bit of it

Edited October 12, 2012 by SoulMonster

[volcano62]

Soul Monster is the fuckin ultimate GNR fan. Fuck all you little juvenile bitches and show some respect.

+ his website/forum is EPIC.

[bbbba]

If you don't know why this is wrong, please go to a mathematician and explain the problem to them. Listen carefully to their approach to the problem and the correct interpretation of data.

Uhm, YOU are the one saying the methodology is wrong, you should be able to explain why it is wrong . So again: Why won't the methodology outlined in the first post succeed at giving numerical values for the amount of similarity between two setlists? The methodology is remarkably simple, it should be very easy to point out what is wrong with it rather than just say it is without value and that another method is better Come on now, stop wasting my time!

The simple version is: because you are pulling numbers and reasons for the numbers out of your ass.

[bionic56]

Soul Monster is the fuckin ultimate GNR fan. Fuck all you little juvenile bitches and show some respect.

+ his website/forum is EPIC.

Volcano marry him, you both have alot in common. The ultimate match and MYGNR's first gay marriage. Respect !

[volcano62]

Soul Monster is the fuckin ultimate GNR fan. Fuck all you little juvenile bitches and show some respect.

+ his website/forum is EPIC.

Volcano marry him, you both have alot in common. The ultimate match and MYGNR's first gay marriage. Respect !

How fuckin' mature...go to hell you littke puke. I'm a fuckin' Christian your gay shit offends me.

At this point i don't give a fuck about you cupcakes or mods suspending/banning me.

If mods don't have the balls to deal with you idiots then i'll put you little bitches in your place

Edited October 12, 2012 by volcano62

[SoulMonster]

If you don't know why this is wrong, please go to a mathematician and explain the problem to them. Listen carefully to their approach to the problem and the correct interpretation of data.

Uhm, YOU are the one saying the methodology is wrong, you should be able to explain why it is wrong . So again: Why won't the methodology outlined in the first post succeed at giving numerical values for the amount of similarity between two setlists? The methodology is remarkably simple, it should be very easy to point out what is wrong with it rather than just say it is without value and that another method is better Come on now, stop wasting my time!

The simple version is: because you are pulling numbers and reasons for the numbers out of your ass.

Uhm, I think you have to give me the not so simplified version because that seemed more like an insult from a person with his back against the wall

Here's my advice to you: Read the first post in this thread, make sure you actually understand the methodology and what it is meant to do. Then make up some fantasy lists (some very similar, some identical, some very dissimilar) and calculate the respective similarity scores between two and two of these lists. See for yourself that the scores correspond to the amount of similarity between the lists. Then come back and continue this discussion.

[bionic56]

Soul Monster is the fuckin ultimate GNR fan. Fuck all you little juvenile bitches and show some respect.

+ his website/forum is EPIC.

Volcano marry him, you both have alot in common. The ultimate match and MYGNR's first gay marriage. Respect !

How fuckin' mature...go to hell you littke puke.

At this point i don't give a fuck about you cupcakes or mods suspending/banning me.

If mods don't have the balls to deal with you idiots then i'll put you little bitches in your place

Ha ! Nice knowing you then, and really you could not put me in my place Volcano, i have the Intellect above a 5 year old so it cuts you out Sir.

[volcano62]

Soul Monster is the fuckin ultimate GNR fan. Fuck all you little juvenile bitches and show some respect.

+ his website/forum is EPIC.

Volcano marry him, you both have alot in common. The ultimate match and MYGNR's first gay marriage. Respect !

How fuckin' mature...go to hell you littke puke.

At this point i don't give a fuck about you cupcakes or mods suspending/banning me.

If mods don't have the balls to deal with you idiots then i'll put you little bitches in your place

Ha ! Nice knowing you then, and really you could not put me in my place Volcano, i have the Intellect above a 5 year old so it cuts you out Sir.

LOL HAHAHAHA you think I give a fuck about your fuckin' ass or this GNR cupcake site? LOLLLLLLL

[bbbba]

If you don't know why this is wrong, please go to a mathematician and explain the problem to them. Listen carefully to their approach to the problem and the correct interpretation of data.

Uhm, YOU are the one saying the methodology is wrong, you should be able to explain why it is wrong . So again: Why won't the methodology outlined in the first post succeed at giving numerical values for the amount of similarity between two setlists? The methodology is remarkably simple, it should be very easy to point out what is wrong with it rather than just say it is without value and that another method is better Come on now, stop wasting my time!

The simple version is: because you are pulling numbers and reasons for the numbers out of your ass.

Uhm, I think you have to give me the not so simplified version because that seemed more like an insult from a person with his back against the wall

Here's my advice to you: Read the first post in this thread, make sure you actually understand the methodology and what it is meant to do. Then make up some fantasy lists (some very similar, some identical, some very dissimilar) and calculate the respective similarity scores between two and two of these lists. See for yourself that the scores correspond to the amount of similarity between the lists. Then come back and continue this discussion.

Similarity score doesn't sound like a proper term. You have to concede here, because you can't justify the premise of ignoring probability. This is not an alternative viewpoint, this is the only viewpoint. Set theory can also be used, but as an adjunct to this approach. Arbitrary scoring and weighting has no basis whatsoever, in the solution of these simple problems.

[bionic56]

Soul Monster is the fuckin ultimate GNR fan. Fuck all you little juvenile bitches and show some respect.

+ his website/forum is EPIC.

Volcano marry him, you both have alot in common. The ultimate match and MYGNR's first gay marriage. Respect !

How fuckin' mature...go to hell you littke puke.

At this point i don't give a fuck about you cupcakes or mods suspending/banning me.

If mods don't have the balls to deal with you idiots then i'll put you little bitches in your place

Ha ! Nice knowing you then, and really you could not put me in my place Volcano, i have the Intellect above a 5 year old so it cuts you out Sir.

LOL HAHAHAHA you think I give a fuck about your fuckin' ass or this GNR cupcake site? LOLLLLLLL

haha be cool man, you' re a great friend of ours.

[volcano62]

Soul Monster is the fuckin ultimate GNR fan. Fuck all you little juvenile bitches and show some respect.

+ his website/forum is EPIC.

Volcano marry him, you both have alot in common. The ultimate match and MYGNR's first gay marriage. Respect !

How fuckin' mature...go to hell you littke puke.

At this point i don't give a fuck about you cupcakes or mods suspending/banning me.

If mods don't have the balls to deal with you idiots then i'll put you little bitches in your place

Ha ! Nice knowing you then, and really you could not put me in my place Volcano, i have the Intellect above a 5 year old so it cuts you out Sir.

LOL HAHAHAHA you think I give a fuck about your fuckin' ass or this GNR cupcake site? LOLLLLLLL

haha be cool man, you' re a great friend of ours.

[bbbba]

Why don't you use the conventional probabilistic approach to analyzing the data, and then we can look at the results? Since you have all the data there, you can start by creating a histogram. Calculate the total the number of times each track (Tr) has ever been played and divide by the total number of tracks ever played. This is the probability of this track being played in any given show P(Tr). Do this for all the tracks and plot P(Tr) values versus the track represented on the dependent axis.

[SoulMonster]

Similarity score doesn't sound like a proper term. You have to concede here, because you can't justify the premise of ignoring probability. This is not an alternative viewpoint, this is the only viewpoint. Set theory can also be used, but as an adjunct to this approach. Arbitrary scoring and weighting has no basis whatsoever, in the solution of these simple problems.

You are just repeating that it isn't the right method without explaining how my method fails at giving scores that correspond to similarities . I suppose it is tempting, if the only tool you have is a hammer, to treat everything as if it were a nail.

I asked you to test out the methodology for yourself, you obviously didn't, but since some other readers might actually still believe you are right, I will give a few easy examples of similarity scores on purely theoretical lists just to show that the methodology works fine for quantifying similarities between lists:

List 1: A, B, C, D, E.

List 2, A, B, C, E, D.

Using my methodology the similarity score is 4.5.

List 1: A, B, C, D, E.

List 3: E, D, C, B, A.

Using my methodology the similarity score is 2.5

List 1: A, B, C, D, E.

List 4: A, B, F, G, H.

Using my methodology the similarity score is 2.0

List 1: A, B; C, D, E.

List 5: F, G, H, I, A.

Using my methodology the similarity score is 0.5.

So answer this simple question: Does not the methodology succeed at giving numerical values to the differences between lists? Are you claiming that the methodology is wrong because the scores are not correlated at all with the amount of similarities? Come on now, this is your time to shine or to fold, stop just saying another method should be used without arguing for why this method is -- in your words -- valueless.

Edited October 12, 2012 by SoulMonster

[bbbba]

Similarity score doesn't sound like a proper term. You have to concede here, because you can't justify the premise of ignoring probability. This is not an alternative viewpoint, this is the only viewpoint. Set theory can also be used, but as an adjunct to this approach. Arbitrary scoring and weighting has no basis whatsoever, in the solution of these simple problems.

You are just repeating that it isn't the right method without explaining how my method fails at giving scores that correspond to similarities . I suppose it is tempting, if the only tool you have is a hammer, to treat everything as if it were a nail.

I asked you to test out the methodology for yourself, you obviously didn't, but since some other readers might actually still believe you are right, I will give a few easy examples of similarity scores on purely theoretical lists just to show that the methodology works fine for quantifying similarities between lists:

List 1: A, B, C, D, E.

List 2, A, B, C, E, D.

Using my methodology the similarity score is 4.5.

List 1: A, B, C, D, E.

List 3: E, D, C, B, A.

Using my methodology the similarity score is 2.5

List 1: A, B, C, D, E.

List 4: A, B, F, G, H.

Using my methodology the similarity score is 2.0

List 1: A, B; C, D, E.

List 5: F, G, H, I, A.

Using my methodology the similarity score is 0.5.

So answer this simple question: Does not the methodology succeed at giving numerical values to the differences between lists? Are you claiming that the methodology is wrong because the scores are not correlated at all with the amount of similarities? Come on now, this is your time to shine or to fold, stop just saying another method should be used without arguing for why this method is -- in your words -- valueless.

Well it's obvious that it doesn't work because any results have to encompass the complete sample set (all shows). Again, just use the tried and trusted method that has been used since the time of the Greeks and we'll be fine.

Similarity score doesn't sound like a proper term. You have to concede here, because you can't justify the premise of ignoring probability. This is not an alternative viewpoint, this is the only viewpoint. Set theory can also be used, but as an adjunct to this approach. Arbitrary scoring and weighting has no basis whatsoever, in the solution of these simple problems.

You are just repeating that it isn't the right method without explaining how my method fails at giving scores that correspond to similarities . I suppose it is tempting, if the only tool you have is a hammer, to treat everything as if it were a nail.

I asked you to test out the methodology for yourself, you obviously didn't, but since some other readers might actually still believe you are right, I will give a few easy examples of similarity scores on purely theoretical lists just to show that the methodology works fine for quantifying similarities between lists:

List 1: A, B, C, D, E.

List 2, A, B, C, E, D.

Using my methodology the similarity score is 4.5.

List 1: A, B, C, D, E.

List 3: E, D, C, B, A.

Using my methodology the similarity score is 2.5

List 1: A, B, C, D, E.

List 4: A, B, F, G, H.

Using my methodology the similarity score is 2.0

List 1: A, B; C, D, E.

List 5: F, G, H, I, A.

Using my methodology the similarity score is 0.5.

So answer this simple question: Does not the methodology succeed at giving numerical values to the differences between lists? Are you claiming that the methodology is wrong because the scores are not correlated at all with the amount of similarities? Come on now, this is your time to shine or to fold, stop just saying another method should be used without arguing for why this method is -- in your words -- valueless.

Well it's obvious that it doesn't work because any results have to encompass the complete sample set (all shows). Again, just use the tried and trusted method that has been used since the time of the Greeks and we'll be fine.

I have an even better idea. Use a neural network e.g. in Matlab to supply all this data. Then we can use the network to predict the Vegas show set lists.

[SoulMonster]

Similarity score doesn't sound like a proper term. You have to concede here, because you can't justify the premise of ignoring probability. This is not an alternative viewpoint, this is the only viewpoint. Set theory can also be used, but as an adjunct to this approach. Arbitrary scoring and weighting has no basis whatsoever, in the solution of these simple problems.

You are just repeating that it isn't the right method without explaining how my method fails at giving scores that correspond to similarities . I suppose it is tempting, if the only tool you have is a hammer, to treat everything as if it were a nail.

I asked you to test out the methodology for yourself, you obviously didn't, but since some other readers might actually still believe you are right, I will give a few easy examples of similarity scores on purely theoretical lists just to show that the methodology works fine for quantifying similarities between lists:

List 1: A, B, C, D, E.

List 2, A, B, C, E, D.

Using my methodology the similarity score is 4.5.

List 1: A, B, C, D, E.

List 3: E, D, C, B, A.

Using my methodology the similarity score is 2.5

List 1: A, B, C, D, E.

List 4: A, B, F, G, H.

Using my methodology the similarity score is 2.0

List 1: A, B; C, D, E.

List 5: F, G, H, I, A.

Using my methodology the similarity score is 0.5.

So answer this simple question: Does not the methodology succeed at giving numerical values to the differences between lists? Are you claiming that the methodology is wrong because the scores are not correlated at all with the amount of similarities? Come on now, this is your time to shine or to fold, stop just saying another method should be used without arguing for why this method is -- in your words -- valueless.

Well it's obvious that it doesn't work

Great! If it is obvious to you then I am sure you can point out how the similarity scores above fail to represent the variations in the setlists. Shouldn't list pair 1-2 have the highest score? Shouldn't list pair 1-5 have the lowest score? Come on, this should be easy, after all the method is without value meaning that there shouldn't be ANY correlation between score and variation.

I have an even better idea. Use a neural network e.g. in Matlab to supply all this data. Then we can use the network to predict the Vegas show set lists.

Haha, yes, using probability theory to determine the probabilities of something happening is sensible. But -- and I have said this a few times now -- we aren't trying to predict any show lists, we are discussing methods for quantifying the amount of variation in two lists.

[bbbba]

Similarity score doesn't sound like a proper term. You have to concede here, because you can't justify the premise of ignoring probability. This is not an alternative viewpoint, this is the only viewpoint. Set theory can also be used, but as an adjunct to this approach. Arbitrary scoring and weighting has no basis whatsoever, in the solution of these simple problems.

You are just repeating that it isn't the right method without explaining how my method fails at giving scores that correspond to similarities . I suppose it is tempting, if the only tool you have is a hammer, to treat everything as if it were a nail.

I asked you to test out the methodology for yourself, you obviously didn't, but since some other readers might actually still believe you are right, I will give a few easy examples of similarity scores on purely theoretical lists just to show that the methodology works fine for quantifying similarities between lists:

List 1: A, B, C, D, E.

List 2, A, B, C, E, D.

Using my methodology the similarity score is 4.5.

List 1: A, B, C, D, E.

List 3: E, D, C, B, A.

Using my methodology the similarity score is 2.5

List 1: A, B, C, D, E.

List 4: A, B, F, G, H.

Using my methodology the similarity score is 2.0

List 1: A, B; C, D, E.

List 5: F, G, H, I, A.

Using my methodology the similarity score is 0.5.

So answer this simple question: Does not the methodology succeed at giving numerical values to the differences between lists? Are you claiming that the methodology is wrong because the scores are not correlated at all with the amount of similarities? Come on now, this is your time to shine or to fold, stop just saying another method should be used without arguing for why this method is -- in your words -- valueless.

Well it's obvious that it doesn't work

Great! If it is obvious to you then I am sure you can point out how the similarity scores above fail to represent the variations in the setlists. Shouldn't list pair 1-2 have the highest score? Shouldn't list pair 1-5 have the lowest score? Come on, this should be easy, after all the method is without value meaning that there shouldn't be ANY correlation between score and variation.

I have an even better idea. Use a neural network e.g. in Matlab to supply all this data. Then we can use the network to predict the Vegas show set lists.

Haha, yes, using probability theory to determine the probabilities of something happening is sensible. But -- and I have said this a few times now -- we aren't trying to predict any show lists, we are discussing methods for quantifying the amount of variation in two lists.

I answer you, but you don't appear to read beyond the first few words. Is this some form of attrition tactic? Anyway, AGAIN, if you avoid using conventional maths to solve an easy problem like this, then you won't get the right answers. If you don't know the conventional method, then ask someone or read up about it.

Edited October 12, 2012 by bbbba