Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss
Log CDF and PDF
12 views
Skip to first unread message
Ilco Petrov
May 20, 2013, 5:21:07 AM5/20/13
to
Dear friends,
%I have defined some log regression with follolwing CDF:
p=[0.001,0.01:0.01:1];
b1=0.01;
b2=0.5;
m=2;
bci2100=p./(1+(((log(p)/log(b1)+log(p)/log(b2)))./m).^m);
plot(p,bci2100)
% 0<p,bci, b1-2<=1
How can I define in Matlab:
- PDF for bci with a.m. dicrete'p' sequence;
- CDF and PDF for bci as continiuos function;
- generate random numbers for 'p' (say sets of 50, 100, 200 for 'p') in (0:1)
- generate a detailed descriptive statistics report real or random 'bci'(similar to Excel and including mean, median, mode, stnandard deviation, sigma, standard mistakes, t-ststistics, hypothesis evaluation,) ; also relative and live reliblility
- Compare 'bci' to a normal distribution and model 'bci' parmeters (b,m) to influence Mu and sigma (location and scale parameters) so than bci to be comparable to a normal distribution.
Please send comments to my email.
Best regards
Ilian Petrov
%I have defined some log regression with follolwing CDF:
p=[0.001,0.01:0.01:1];
b1=0.01;
b2=0.5;
m=2;
bci2100=p./(1+(((log(p)/log(b1)+log(p)/log(b2)))./m).^m);
plot(p,bci2100)
% 0<p,bci, b1-2<=1
How can I define in Matlab:
- PDF for bci with a.m. dicrete'p' sequence;
- CDF and PDF for bci as continiuos function;
- generate random numbers for 'p' (say sets of 50, 100, 200 for 'p') in (0:1)
- generate a detailed descriptive statistics report real or random 'bci'(similar to Excel and including mean, median, mode, stnandard deviation, sigma, standard mistakes, t-ststistics, hypothesis evaluation,) ; also relative and live reliblility
- Compare 'bci' to a normal distribution and model 'bci' parmeters (b,m) to influence Mu and sigma (location and scale parameters) so than bci to be comparable to a normal distribution.
Please send comments to my email.
Best regards
Ilian Petrov
TideMan
May 20, 2013, 6:10:14 AM5/20/13
to
What have you done towards solving the problem?
Show us your code and where you're having trouble, then someone might help.
Or would you prefer us to just give you the code?
Then you could claim our work as your own.
The English word for people who do that is: CHEAT.
Ilco Petrov
May 20, 2013, 8:06:09 AM5/20/13
to
TideMan <mul...@gmail.com> wrote in message <14d25051-36d7-4458...@googlegroups.com>...
Dear Sir (Mr TiedMan),
Thanks for being very quick, kind and polite in your own way!
I have all thinks in in Excel, where I did not have any problem! My publication is already published in a top oil and gas magazine!
Sorry, I am new to Matlab, as I started to copy my project to another package since it will be a long project and it is better to recheck resullts. And to see if Matlab is better as product and service support. So far the second is not obvious at all, but I guess you are not a typical case, rather an...
It is not at all student homewok as I am a phd from Bulgaria (EU) based here in Moscow, where there are not much publications for Matlab, in Bulgaria also.
My colleages here in Moscow are very kind and the Matlab community is very supportive and cooperative. But they work on technical aspects, while my project is in economics and there is no expertise in logistics distributions. The point is that my aproach to the log regression is specific and does not fit with typical ways of expression and formula.
The questions I asked are routine - I am not asking you a code for a modelq Or some top secret data or individual algorithm.
If you have some publications and similar scrip, please simple advise - do not insult me!
I looked in your Distribution fitting tool, but there I can not load my formula, just data - I do not need that.
I already sent you my code for the bci, which is actually not 'vanila' log regression, but have some elements of neuron network even at this level.
If you do not have experience in log regressions, better do not be that agressive - you are talking to real persons and clients. The questions I asked are routine - I am not asking you a code for a model - just simple thinksabout simple technical instruments. And Matlab is not a unique progarm. You are not unique, also.
As far as I understand your job is to help people, not to insult them hiding behind the virtulal curtain and thinking you are untuchable. Giving technical saaistance is not giving godness
My self, I am a lawyer, also.
So, kindly I ask you to the the following:
- to make an official excuse to me personally to me and all readers for your impertinent expression, as this blog is public, but not yours and you are not at home. You are paid, I assume, to help custummers, not to insult them.
- to pass my questions to somebody who is more polite and cooperative and our corespondence to your boss for opinion and comment.
Only an excuse will stop me develop a legal case in due time.
Expecting immediate excuse from you and/or your boss.
Ilian Petrov
Thanks for being very quick, kind and polite in your own way!
I have all thinks in in Excel, where I did not have any problem! My publication is already published in a top oil and gas magazine!
Sorry, I am new to Matlab, as I started to copy my project to another package since it will be a long project and it is better to recheck resullts. And to see if Matlab is better as product and service support. So far the second is not obvious at all, but I guess you are not a typical case, rather an...
It is not at all student homewok as I am a phd from Bulgaria (EU) based here in Moscow, where there are not much publications for Matlab, in Bulgaria also.
My colleages here in Moscow are very kind and the Matlab community is very supportive and cooperative. But they work on technical aspects, while my project is in economics and there is no expertise in logistics distributions. The point is that my aproach to the log regression is specific and does not fit with typical ways of expression and formula.
The questions I asked are routine - I am not asking you a code for a modelq Or some top secret data or individual algorithm.
If you have some publications and similar scrip, please simple advise - do not insult me!
I looked in your Distribution fitting tool, but there I can not load my formula, just data - I do not need that.
I already sent you my code for the bci, which is actually not 'vanila' log regression, but have some elements of neuron network even at this level.
If you do not have experience in log regressions, better do not be that agressive - you are talking to real persons and clients. The questions I asked are routine - I am not asking you a code for a model - just simple thinksabout simple technical instruments. And Matlab is not a unique progarm. You are not unique, also.
As far as I understand your job is to help people, not to insult them hiding behind the virtulal curtain and thinking you are untuchable. Giving technical saaistance is not giving godness
My self, I am a lawyer, also.
So, kindly I ask you to the the following:
- to make an official excuse to me personally to me and all readers for your impertinent expression, as this blog is public, but not yours and you are not at home. You are paid, I assume, to help custummers, not to insult them.
- to pass my questions to somebody who is more polite and cooperative and our corespondence to your boss for opinion and comment.
Only an excuse will stop me develop a legal case in due time.
Expecting immediate excuse from you and/or your boss.
Ilian Petrov
Ilco Petrov
May 20, 2013, 8:48:08 AM5/20/13
to
Bjorn Gustavsson
May 20, 2013, 11:15:09 AM5/20/13
to
"Ilco Petrov" <ili...@mail.ru> wrote in message <knd3jh$cm$1...@newscl01ah.mathworks.com>...
Let us just cut this nonsense short right here. This is a internet newsgroup with voluntary contributions from randomly distributed (geographically, professionally and so on) people, who might be working with technical assistance - but the customer assistance from people working for Mathworks is not through this channel. If IIRC Tideman works in oceanography, but that might be a complete misunderstanding of mine. There are a fair number of homework type messages requesting solutions to this newsgroup and your message had the same look to it. As a lawyer you should have the schooling to learn what the rules of the forum you venture in are, shouldn't you? Had you done that you would have made your message clearly stated in such a way that it was obvious for everyone that it was not a homework...
HTH
HTH
Ilco Petrov
May 20, 2013, 2:38:08 PM5/20/13
to
"Bjorn Gustavsson" <bj...@irf.se> wrote in message <kndelt$928$1...@newscl01ah.mathworks.com>...
> Bjorn: I thoght this blog is supported by Matworks for the assitance of Matlab users. I could not imagine such impertinent behavior in a community pretending to be hilgly intelectual and educated. I realize now that this blog is full of nice people, who ask very direct interesting question and seldom receive good and specific answers - Most often the answers are simililar to "go to some item in help".
This supports my impression, that Matlab is a very good package, but the environment is less user friendly than others - otherwise users would learn its basics quickly and not ask so much simple routine qestions, that may be dull for gurus (like tidmans). In education and science the experiment and the methods should be priority - not the tools. The tools should be clear and easy to use. if you need to spend more time to write every script from A to Z than on developing method and experiment - too bad... From economics point of view it is linked relation with clients and users (clients already).
Still I do not see any real gurus in Logistics distrubutions and logit regressions in economics here, that may provide sound expertise and advise. This is not only our (users) problem ... this may be regarded as a problem of Matlab and Matwork.
Your post does not add value - rather seems to some mentor's talk combined with multi conditional statements about something you are not specialistl in (the legal profession). I see you are working in very distant area from economics (and law). For an economist it was clear from the beginning, that my questions are not for "homework". I cannot make such difference in your area - and if I am not not specialist in some are, I never try to be the "salt & pepper for every dish". This is a matter of home - not academic education!
This supports my impression, that Matlab is a very good package, but the environment is less user friendly than others - otherwise users would learn its basics quickly and not ask so much simple routine qestions, that may be dull for gurus (like tidmans). In education and science the experiment and the methods should be priority - not the tools. The tools should be clear and easy to use. if you need to spend more time to write every script from A to Z than on developing method and experiment - too bad... From economics point of view it is linked relation with clients and users (clients already).
Still I do not see any real gurus in Logistics distrubutions and logit regressions in economics here, that may provide sound expertise and advise. This is not only our (users) problem ... this may be regarded as a problem of Matlab and Matwork.
Your post does not add value - rather seems to some mentor's talk combined with multi conditional statements about something you are not specialistl in (the legal profession). I see you are working in very distant area from economics (and law). For an economist it was clear from the beginning, that my questions are not for "homework". I cannot make such difference in your area - and if I am not not specialist in some are, I never try to be the "salt & pepper for every dish". This is a matter of home - not academic education!
Bjorn Gustavsson
May 20, 2013, 6:07:09 PM5/20/13
to
"Ilco Petrov" <ili...@mail.ru> wrote in message <kndqig$iua$1...@newscl01ah.mathworks.com>...
Well, impertinent, indeed: impertinent.
In my field of work I haven't run into logistic regression problems, so I can only refer you to the documentation for mnrfit, and mnrval.
In my field of work I haven't run into logistic regression problems, so I can only refer you to the documentation for mnrfit, and mnrval.
>
> This supports my impression, that Matlab is a very good package, but the environment is less user friendly than others - otherwise users would learn its basics quickly and not ask so much simple routine qestions, that may be dull for gurus (like tidmans). In
>
I think you underestimate the vast number of students that face matlab for the first time every semester, and the number of students that has home-work assignments in various courses from algebra, statistics, physics economics and whatnot.
> This supports my impression, that Matlab is a very good package, but the environment is less user friendly than others - otherwise users would learn its basics quickly and not ask so much simple routine qestions, that may be dull for gurus (like tidmans). In
>
>
education and science the experiment and the methods should be priority - not the tools. The tools should be clear and easy to use. if you need to spend more time to write every script from A to Z than on developing method and experiment - too bad... From
>
Tools and methods has _always_ been developed and made by someone, to have new and improved tools in the future, education has to teach some how such tools are developed and made. That you have the fortune to see this in a slightly different light is fortunate for you.
education and science the experiment and the methods should be priority - not the tools. The tools should be clear and easy to use. if you need to spend more time to write every script from A to Z than on developing method and experiment - too bad... From
>
>
economics point of view it is linked relation with clients and users (clients already).
> Still I do not see any real gurus in Logistics distrubutions and logit regressions in economics here, that may provide sound expertise and advise. This is not only our (users) problem ... this may be regarded as a problem of Matlab and Matwork.
>
> Your post does not add value - rather seems to some mentor's talk combined with multi conditional statements about something you are not specialistl in (the legal profession).
>
I actually thought it to be a bit more basic than that. But lets skip manners.
economics point of view it is linked relation with clients and users (clients already).
> Still I do not see any real gurus in Logistics distrubutions and logit regressions in economics here, that may provide sound expertise and advise. This is not only our (users) problem ... this may be regarded as a problem of Matlab and Matwork.
>
> Your post does not add value - rather seems to some mentor's talk combined with multi conditional statements about something you are not specialistl in (the legal profession).
>
>
I see you are working in very distant area from economics (and law). For an economist it was clear from the beginning, that my questions are not for "homework". I cannot make such difference in your area - and if I am not not specialist in some are, I never try to be the "salt & pepper for every dish". This is a matter of home - not academic education!
For Tideman and me your question seemed to be perfectly reasonable for a course in statistics, so there's that.
I see you are working in very distant area from economics (and law). For an economist it was clear from the beginning, that my questions are not for "homework". I cannot make such difference in your area - and if I am not not specialist in some are, I never try to be the "salt & pepper for every dish". This is a matter of home - not academic education!
Tom Lane
May 21, 2013, 6:27:52 PM5/21/13
to
> bci2100=p./(1+(((log(p)/log(b1)+log(p)/log(b2)))./m).^m);
> plot(p,bci2100)
> % 0<p,bci, b1-2<=1
>
> How can I define in Matlab:
> - PDF for bci with a.m. dicrete'p' sequence;
> - CDF and PDF for bci as continiuos function;
> ...
> plot(p,bci2100)
> % 0<p,bci, b1-2<=1
>
> How can I define in Matlab:
> - PDF for bci with a.m. dicrete'p' sequence;
> - CDF and PDF for bci as continiuos function;
You have quite a debate going on with other folks.
I would be interesting in answering your question, but I don't understand
it. If you want to generate random p values and use the bci2100 expression
to derive random bci2100 values, this does that:
F = @(p) p./(1+(((log(p)/log(b1)+log(p)/log(b2)))./m).^m);
newbci = F(rand(1000,1));
hist(newbci, 40)
If you want to treat your curve as a cdf and invert it, in other words
generate random bci2100 values and find the corresponding p values, you
could do the following interpolation as an approximation:
plot(p,bci2100)
pnew = interp1(bci2100,p,rand(1000,1));
hold on; ecdf(pnew); hold off % <-- for comparison, if you have the
Statistics Toolbox
It's unlikely you will find someone to "generate a detailed descriptive
statistics report" for you, but if you ask a specific question related to
that you may get some help.
-- Tom
0 new messages