how to solve this higher polynomial function?
in [Matlab]
|
From: bo on 4 Mar 2010 11:02 dpb <none(a)non.net> wrote in message <hmokmh$ub9$1(a)news.eternal-september.org>... > bo wrote: > > "John D'Errico" <woodchips(a)rochester.rr.com> wrote in message > > <hmljbd$jjt$1(a)fred.mathworks.com>... > > ...[a very nice commentary on regression modeling elided only for > brevity]... > > > Thanks a lot John.But tell you I have only learnt matlab 2 weeks. Have > > not learnt how to walk but to run already.No choice it is my > > project.Anyway, thanks for your time and patience. > > > This is _NOT_ a Matlab issue John is addressing. It is the fundamentals > of the problem and the methodology itself. Unless and until you follow > the esteemed John D'E's advice and understand the bases and meaning of > regression modeling and what can (and probably even more importantly > canNOT) be inferred your project is doomed to failure. > > Start w/ a tutorial on regression and/or other modeling techniques. If > you must, find the consulting statistics group in your university and > make use of them to at least get pointers to appropriate texts. > > -- Thanks both of you.It is my fault to address problem like this.I think I will study some tutorials regarding regression first before I really come to this problem.So sorry about that.
From: John D'Errico on 4 Mar 2010 11:45 "bo " <bobpong1979(a)hotmail.com> wrote in message <hmolhu$esr$1(a)fred.mathworks.com>... > dpb <none(a)non.net> wrote in message <hmokmh$ub9$1(a)news.eternal-september.org>... > > bo wrote: > > > "John D'Errico" <woodchips(a)rochester.rr.com> wrote in message > > > <hmljbd$jjt$1(a)fred.mathworks.com>... > > > > ...[a very nice commentary on regression modeling elided only for > > brevity]... > > > > > Thanks a lot John.But tell you I have only learnt matlab 2 weeks. Have > > > not learnt how to walk but to run already.No choice it is my > > > project.Anyway, thanks for your time and patience. > > > > > > This is _NOT_ a Matlab issue John is addressing. It is the fundamentals > > of the problem and the methodology itself. Unless and until you follow > > the esteemed John D'E's advice and understand the bases and meaning of > > regression modeling and what can (and probably even more importantly > > canNOT) be inferred your project is doomed to failure. > > > > Start w/ a tutorial on regression and/or other modeling techniques. If > > you must, find the consulting statistics group in your university and > > make use of them to at least get pointers to appropriate texts. > > > > -- > > Thanks both of you.It is my fault to address problem like this.I think I will study some tutorials regarding regression first before I really come to this problem.So sorry about that. As we have said, you need to use a lower order model in this regression. There is a big tendency among novice users of polynomial regression to decide that 1. The low order model looks nice, but is not accurate enough. 2. Using a higher order model fits the data with lower errors, therefore it must be better. They invariably come to the conclusion that impossibly high order models must be the best, the higher the order the better, until MATLAB finally just gives up and refuses to fit the model they have posed. The computer will do whatever you tell it to do, with no complaints. Use a lower order model, or recognize that your data is insufficient to fit the model you have chosen. In this event, you might choose to get better data. John
From: bo on 4 Mar 2010 13:06 "John D'Errico" <woodchips(a)rochester.rr.com> wrote in message <hmoo2v$1oo$1(a)fred.mathworks.com>... > "bo " <bobpong1979(a)hotmail.com> wrote in message <hmolhu$esr$1(a)fred.mathworks.com>... > > dpb <none(a)non.net> wrote in message <hmokmh$ub9$1(a)news.eternal-september.org>... > > > bo wrote: > > > > "John D'Errico" <woodchips(a)rochester.rr.com> wrote in message > > > > <hmljbd$jjt$1(a)fred.mathworks.com>... > > > > > > ...[a very nice commentary on regression modeling elided only for > > > brevity]... > > > > > > > Thanks a lot John.But tell you I have only learnt matlab 2 weeks. Have > > > > not learnt how to walk but to run already.No choice it is my > > > > project.Anyway, thanks for your time and patience. > > > > > > > > > This is _NOT_ a Matlab issue John is addressing. It is the fundamentals > > > of the problem and the methodology itself. Unless and until you follow > > > the esteemed John D'E's advice and understand the bases and meaning of > > > regression modeling and what can (and probably even more importantly > > > canNOT) be inferred your project is doomed to failure. > > > > > > Start w/ a tutorial on regression and/or other modeling techniques. If > > > you must, find the consulting statistics group in your university and > > > make use of them to at least get pointers to appropriate texts. > > > > > > -- > > > > Thanks both of you.It is my fault to address problem like this.I think I will study some tutorials regarding regression first before I really come to this problem.So sorry about that. > > As we have said, you need to use a lower order model > in this regression. There is a big tendency among novice > users of polynomial regression to decide that > > 1. The low order model looks nice, but is not accurate > enough. > > 2. Using a higher order model fits the data with lower > errors, therefore it must be better. > > They invariably come to the conclusion that impossibly > high order models must be the best, the higher the order > the better, until MATLAB finally just gives up and refuses > to fit the model they have posed. The computer will do > whatever you tell it to do, with no complaints. > > Use a lower order model, or recognize that your data is > insufficient to fit the model you have chosen. In this > event, you might choose to get better data. > > John Ok.Roger that.I will try to use a regression model to project or predict the trend curve passing through those data points .If more points are added in subsequently, this curve will also show the trend.Is it ok? bo
From: dpb on 4 Mar 2010 13:31 bo wrote: > "John D'Errico" <woodchips(a)rochester.rr.com> wrote in message > <hmoo2v$1oo$1(a)fred.mathworks.com>... .... >> Use a lower order model, or recognize that your data is >> insufficient to fit the model you have chosen. In this >> event, you might choose to get better data. >> .... > Ok.Roger that.I will try to use a regression model to project or predict > the trend curve passing through those data points .If more points are > added in subsequently, this curve will also show the trend.Is it ok? Better, anyway... :) Again, can't overemphasize the need to look at some underlying theory to get a handle on what the statistics mean (and the assumptions made on the data behind deriving them) in order to make reasonable judgments of whether any of it makes any sense at all. Certainly as John showed, at least looking at error bands on estimated coefficients is a start. For regression before ever even considering a model I can't over emphasize the desirability of looking at plots of the data initially as well as consideration of the location of the independent variables in n-space, etc., etc., etc., ... Again, this is the sort of think your advisor/prof/whoever should be able to provide some guidance on before you get too far gone in just data fitting. --
From: bo on 4 Mar 2010 14:10 dpb <none(a)non.net> wrote in message <hmoudf$pd$1(a)news.eternal-september.org>... > bo wrote: > > "John D'Errico" <woodchips(a)rochester.rr.com> wrote in message > > <hmoo2v$1oo$1(a)fred.mathworks.com>... > ... > > >> Use a lower order model, or recognize that your data is > >> insufficient to fit the model you have chosen. In this > >> event, you might choose to get better data. > >> > ... > > > Ok.Roger that.I will try to use a regression model to project or predict > > the trend curve passing through those data points .If more points are > > added in subsequently, this curve will also show the trend.Is it ok? > > Better, anyway... :) > > Again, can't overemphasize the need to look at some underlying theory to > get a handle on what the statistics mean (and the assumptions made on > the data behind deriving them) in order to make reasonable judgments of > whether any of it makes any sense at all. > > Certainly as John showed, at least looking at error bands on estimated > coefficients is a start. > > For regression before ever even considering a model I can't over > emphasize the desirability of looking at plots of the data initially as > well as consideration of the location of the independent variables in > n-space, etc., etc., etc., ... > > Again, this is the sort of think your advisor/prof/whoever should be > able to provide some guidance on before you get too far gone in just > data fitting. > > -- You are absolutely correct.Actually when I showed my advisor this uselss higher order polynomial function, as I am also a novice to matlab,my advisor should stop me at first place rather than asking me to solve it.So I just follow whatever my advisor said to do.That is why my direction totally went wrong. Is there any recommendation website of regression modeling? Thanks
First
|
Prev
|
Next
|
Last
Pages: 1 2 3 Prev: C code for simulink model to be used with TMS Next: StatMax - One Weapon for All Calculations |