matlab에서 fitcsvm함수로 SVM분류기를 이용해 ROC curve를 그리려면, 학습한 SVM 모델을 fitPosterior함수(score 를 posterior probability로 변환)를 통해 모델을 변환한 후 predict함수의 입력모델로 써야 해줘야 test셋의 posterior probability를 구할 수 있다. 이 posterior probability를 perfcurve에 입력시켜 roc curve와 auc를 그릴 수 있다. 아래는 그 예제.

test_fitcsvm_predict_perfcurve1.m

close all; clear; clc;

% Load the ionosphere data set. Suppose that the last 10 observations become available after training the SVM classifier.
load ionosphere

n = size(X,1);       % Training sample size
% isInds = 1:(n-10);   % In-sample indices
% oosInds = (n-9):n;   % Out-of-sample indices
cp = cvpartition(Y, 'k', 5);
disp(cp)
trIdx = cp.training(1);
teIdx = cp.test(1);
isInds = find(trIdx);
oosInds = find(teIdx);

% Train an SVM classifier. It is good practice to standardize the predictors and specify the order of the classes.
% Conserve memory by reducing the size of the trained SVM classifier.
SVMModel = fitcsvm(X(isInds,:),Y(isInds),'Standardize',true, 'ClassNames',{'b','g'});
CompactSVMModel = compact(SVMModel);
whos('SVMModel','CompactSVMModel')

% The positive class is 'g'. The CompactClassificationSVM classifier (CompactSVMModel) uses less space than
% the ClassificationSVM classifier (SVMModel) because the latter stores the data.

% Estimate the optimal score-to-posterior-probability-transformation function.
fCompactSVMModel = fitPosterior(CompactSVMModel, X(isInds,:),Y(isInds))

% The optimal score transformation function (CompactSVMModel.ScoreTransform) is the sigmoid function
% because the classes are inseparable.
% Predict the out-of-sample labels and positive class posterior probabilities.
% Since true labels are available, compare them with the predicted labels.

[labels,PostProbs] = predict(fCompactSVMModel,X(oosInds,:));
table(Y(oosInds),labels,PostProbs(:,2),'VariableNames', {'TrueLabels','PredictedLabels','PosClassPosterior'})

a1 = Y(oosInds);
a2 = PostProbs(:,2);
a3 = Y{1};
[falsePositiveTree, truePositiveTree, T, AucTreeg] = perfcurve(Y(oosInds),PostProbs(:,2), Y{1});
plot(falsePositiveTree, truePositiveTree, 'LineWidth', 5)
xlabel('False positive rate');
ylabel('True positive rate');
title('ROC');

% a1 = Y(oosInds);
% a2 = PostProbs(:,2);
% a3 = Y{2};
% figure,
% [falsePositiveTree, truePositiveTree, T, AucTreeb] = perfcurve(Y(oosInds),PostProbs(:,2), Y{2});
% plot(falsePositiveTree, truePositiveTree, 'LineWidth', 5)
% xlabel('False positive rate');
% ylabel('True positive rate');
% title('ROC');

[~,scoresSVMModel] = predict(SVMModel,X(oosInds,:));
[~,scoresCompactSVMModel] = predict(CompactSVMModel,X(oosInds,:));
[~,scoresfCompactSVMModel] = predict(fCompactSVMModel,X(oosInds,:));
[falsePositiveTree, truePositiveTree, T, AucTreeg] = perfcurve(Y(oosInds),scores(:,2), Y{1});
plot(falsePositiveTree, truePositiveTree, 'LineWidth', 5)
xlabel('False positive rate');
ylabel('True positive rate');
title('ROC');

a1 = 1;

https://kr.mathworks.com/matlabcentral/answers/99067-why-do-i-receive-the-error-license-manager-error-9

When I try to launch MATLAB, I get the following error:

 License checkout failed. Invalid host. License Manager Error -9

-----------------------------------------

The two most common causes of License Manager Error -9 are:

• The MATLAB license file was activated for a different computer and/or user account
• You have a Designated Computer MATLAB license and MATLAB is already running under a different user account on this computer

MATLAB was activated for a different Computer and/or User Account

More specifically, this means that the host ID specified in the license file is not the host ID of the computer. If you are using an Individual license, this may also indicate that the login name specified in the license file does not match the logged-in user account.

To resolve this issue, you need to reactivate MATLAB using the correct host ID/Username:

What is a Host ID? How do I find my Host ID in order to activate my license?

https://www.mathworks.com/matlabcentral/answers/101892

How do I find my Login Name in order to activate my license?

https://www.mathworks.com/matlabcentral/answers/96800

How do I activate MATLAB?

http://www.mathworks.com/matlabcentral/answers/99457

If reactivating does not resolve the license manager error -9, and you are using a Designated Computer license, another user might already be using the license under a different account:

MATLAB is running on a different user account

The Designated Computer license option only allows one account on a computer to run MATLAB at a time. If you are using a Designated Computer license and have already confirmed that your host ID is correct, then there is probably an instance of MATLAB running under a different user account on this machine.

To test this, log out any other users on this machine. Be sure to check with your colleague to make sure that none of them have unsaved worked open on those other accounts before logging them out.

Look at the output of,

which -all pca

The first item should be something ending in \toolbox\stats\stats\pca.m. My guess is that you have another pca.m somewhere on your path.

http://stackoverflow.com/questions/19643569/too-many-input-arguments-even-with-varargin

Unfortunately I am getting a "Too many input arguments." error from performing this call:

[varargout{1:nargout}]=pca(varargin{1},'Algorithm','svd','Economy',fEconomy);

on the function that has signature as follows:

function [coeff, score, latent, tsquared, explained, mu] = pca(x,varargin)

I am calling princomp in this way:

[pc,score,latent,tsquare] = princomp(data);

Any idea of what might be the cause? (the bug appears in princomp.m of the stats package)

----------------------------------------------------------

Look at the output of,

which -all pca

The first item should be something ending in \toolbox\stats\stats\pca.m. My guess is that you have another pca.m somewhere on your path.

아래 [1] 주소에 나와있는대로, rbf 커널로 하고, boxconstraint 와 rbf_sigma 를 설정한다. 이때, fitcsvm함수에서의 rbf_sigma값은 'KernelScale'로 설정한다 [2]. [2]에서 The software divides all elements of the predictor matrix X by the value of KernelScale.라고 돼 있는데, 이게 sigma인 이유는 [3]에서 매트릭스를 둘다 나눈 것은 결국 sigma역할을 하므로.

[1] http://www.mathworks.com/help/releases/R2013a/stats/support-vector-machines-svm.html#bsr5o3f

• Start with Kernel_Function set to 'rbf' and default parameters.

• Try different parameters for training, and check via cross validation to obtain the best parameters.

The most important parameters to try changing are:

• boxconstraint — One strategy is to try a geometric sequence of the box constraint parameter. For example, take 11 values, from 1e-5 to 1e5 by a factor of 10.

• rbf_sigma — One strategy is to try a geometric sequence of the RBF sigma parameter. For example, take 11 values, from 1e-5 to 1e5 by a factor of 10.

[2]https://www.mathworks.com/help/stats/fitcsvm.html#bt7oo83-5

'KernelScale' — Kernel scale parameter
1 (default) | 'auto' | positive scalar

[3] https://www.mathworks.com/matlabcentral/newsreader/view_thread/339612

Doc says that for these 3 kernels "the software divides all elements of
the predictor matrix X by the value of KernelScale". The unscaled
Gaussian (aka RBF) kernel is

G(x,z) = exp(-(x-z)'*(x-z))

for column-vectors x and z. If KernelScale is s, its scaled version is

G(x,z) = exp(-(x/s-z/s)'*(x/s-z/s))

or equivalently

G(x,z) = exp(-(x-z)'*(x-z)/s^2)

Setting KernelScale sets the RBF sigma.

-Ilya

[4]http://stackoverflow.com/questions/40226658/change-the-value-of-sigma-using-the-fitcsvm-function-in-matlab

https://www.mathworks.com/matlabcentral/answers/125704-how-can-i-save-a-figure-to-a-tiff-image-with-resolution-300-dpi-in-matlab

1. Using the MATLAB Graphical Environment

Select the figure window that shows the figure to be saved and then follow the steps mentioned below:

• Click on File>>Export Setup.

• A menu named as “Export Setup: Figure Name” must have popped up.

• On the left side of the menu, choose the option named “Rendering”. After that, change the value of:

        - ‘Colorspace’ to ’grayscale’

        - ‘Resolution(dpi)’ to 300

This step is shown in image file “Step3.jpeg” attached to the solution.

• Click on ‘Apply to Figure’ button.

• Click ‘OK’ button.

• The menu should have disappeared and you should see the figure window with all your changes applied.

• Click on File>>Save As.

• Now, you can save the file to your system. Please do not forget to save as “TIFF image”.

2. Saving a figure to an image file programmatically

  a) Use "print" as: 

>> print(gcf, '-dtiff', 'myfigure.tiff'); 

b) Alternatively, you could do: 

>> imagewd = getframe(gcf); 
>> imwrite(imagewd.cdata, 'myfigure.tiff'); 

To change the quality of a TIFF file, you can specify 'Compression',''none' and increase the resolution with 'Resolution',300.

https://www.mathworks.com/matlabcentral/answers/233545-is-there-a-way-to-connect-to-microsoft-sql-server-database-without-needing-the-database-toolbox

Hello,

I was wondering if there is a way to connect to Microsoft SQL Server database without the database toolbox?? I did some research on JDBC and OBDC drivers, but am unsure where to go from there. I can install the drivers and see how far I can get, but before taking this step and doing the setup, I would like to know beforehand if it is possible to connect to Microsoft SQL Server database without the database toolbox. The two following links look like I need database toolbox to connect to it.

http://www.mathworks.com/help/releases/R2015a/database/ug/microsoft-sql-server-jdbc-windows.html

http://www.mathworks.com/help/database/ug/microsoft-sql-server-odbc-windows.html

I am using 2015a btw student edition. Any help and suggestions would be appreciated. Thanks.

------------------------------------------------------

Did you search the File Exchange for SQL Server?

Google found

• adodb_tools by Jaroslaw Tuszynski   "... designed to work on Microsoft SQL Server, Oracle, Microsoft Access, MySQL and other databases."   Comment: "Great little script, much faster than the database toolbox supplied by matlab. Better documentation would be nice though."
• CSE SQL Database Library by Laurent Cavin   "EDITOR'S NOTE: Functionality for data exchange between MATLAB and ODBC/JDBC compliant databases is also provided with the MATLAB Database Toolbox."
• JDBC and Matlab: A simple example   "... . However, you do not need to purchase this (database toolbox) to interact with your database."
• Connect to SQL Server from Matlab

https://www.mathworks.com/matlabcentral/answers/10428-standard-deviation-and-mean

https://www.mathworks.com/help/matlab/ref/errorbar.html

Hello everybody,

I have 36 values of mean and their standard deviation. 12 values falls between 38 to 45, another 12 values falls between 53 to 60 and another 12 values fall between70 to 75. I just want to show in a graph clearly the mean values and their standard deviation. I tried so many but none of them are really clear bcoz, I have values like 43.77, 43.10, 43.5... some close values.. how can I do that.. help me

----------------------------------------------------------

You can use errorbar:

% The data

Y = [rand(12,1)*7 + 35; rand(12,1)*7 + 53; rand(12,1)*5 + 70];

% The standard deviations

E = rand(36,1)*10 + 3;

% Mean values with error bars

errorbar(Y,E,'x')

Suppose that;

I have an m-file at location:
C:\M1\M2\M3\mfile.m

And exe file of the matlab is at this location:
C:\E1\E2\E3\matlab.exe

I want to run this m-file with Matlab, from command-line, for example inside a .bat file. How can I do this, is there a way to do it?

----------------------------------------------------

A command like this runs the m-file successfully:

"C:\<a long path here>\matlab.exe" -nodisplay -nosplash -nodesktop -r "run('C:\<a long path here>\mfile.m');"

https://www.mathworks.com/matlabcentral/answers/102082-how-do-i-call-matlab-from-the-dos-prompt

To start MATLAB from a DOS window running inside Windows, do the following:

1. Open a DOS prompt

2. Change directories to $MATLABROOT\bin

(where $MATLABROOT is the MATLAB root directory on your machine, as returned by typing

matlabroot

at the MATLAB Command Prompt.)

3. Type "matlab"

You can also create your own batch file in the $MATLABROOT\bin directory

NOTE: If you have other commands that follow the call to MATLAB in your batch file, use matlab.exe rather than matlab.bat. If you call on matlab.bat, subsequent commands in the batch file will not get executed.

To do so, use the following instructions:

1. Create a file called mat.bat and place the following line into it:

 win $MATLABROOT\bin\matlab.exe

2. Insert $MATLABROOT\bin into the path in the autoexec.bat file.

(where $MATLABROOT is the MATLAB root directory on your machine, as returned by typing

matlabroot

at the MATLAB Command Prompt.)

Now you can type "mat" at the dos prompt and Windows will come up with MATLAB.

You can run MATLAB from the DOS prompt and save the session to an output file by doing the following:

 matlab -r matlab_filename_here -logfile c:\temp\logfile

Depending on the directory you are in, you may need to specify the path to the executable. The MATLAB file you want to run must be on your path or in the directory. This MATLAB file can be a function that takes arguments or a script.

When running a script 'myfile.m', use the following command:

 matlab -r myfile

When calling a function 'myfile.m' which accepts two arguments:

 matlab -r myfile(arg1,arg2)

To pass numeric values into 'myfile.m' simply replace 'arg1' and 'arg2' with numeric values. To pass string or character values into 'myfile.m' replace 'arg1' and 'arg2' with the string or character values surrounded in single quotes. For exampl to pass the string values 'hello' and 'world' into 'myfile.m' use the following command:

 matlab -r myfile('hello','world')

Note that the logfile will contain everything that was displayed to the Command Window while the MATLAB file was running. If you want to generate any print files you need to do this in the MATLAB file. You can combine this example with the above one to create a batch file that takes input files and creates output files.

In addition, this will call up an additional instance of the MATLAB command window. If you wish this to exit after the computation is complete, you will need to add the command 'exit' to the end of your MATLAB file. You can suppress the splash screen by adding the -nosplash flag to the above command so it looks like the following:

 matlab -nosplash -r mfile -logfile c:\temp\logfile

Although you cannot prevent MATLAB from creating a window when starting on Windows systems, you can force the window to be hidden, by using the start command with the -nodesktop and -minimize options together:

 start matlab -nosplash -nodesktop -minimize -r matlab_filename_here -logfile c:\temp\logfile

If you would like to call multiple MATLAB functions using the -r switch you could write a single function which will call each of the other MATLAB functions in the desired order.

Note: Batch files can be called from Windows scheduler in order to run MATLAB commands at specific times. May not work for UNC pathnames.

Map grayscale to color using colormap

imshow(im)

imshow(im,'Colormap',jet(255))

rgb = ind2rgb(gray2ind(im,255),jet(255));

n = size(unique(reshape(im,size(im,1)*size(im,2),size(im,3))),1);

rgb = label2rgb(gray2ind(im,255),jet(255));

imshow(rgb);

아래 수식에서 p(i,j)는 normalized GLCM, i,j는 0~8 위치를 나타냄.

즉, contrast는 GLCM의 거리에 따라 weight주어 평균 냄. 대각선은 거리가 0이고, 우상,좌하쪽으로 갈수록 거리가 커짐. 대각선에서 우상,좌하쪽으로 갈수록 GLCM은 밝기차가 큼을 나타냄.

%-----------------------------------------------------------------------------
function C = calculateContrast(glcm,r,c)
% Reference: Haralick RM, Shapiro LG. Computer and Robot Vision: Vol. 1,
% Addison-Wesley, 1992, p. 460. 
k = 2;
l = 1;
term1 = abs(r - c).^k;
term2 = glcm.^l;
 
term = term1 .* term2(:);
C = sum(term);

http://mathforum.org/kb/message.jspa?messageID=6998158

graycomatrix가 [0,255] 사이의 이미지 입력 시 [0,8]으로 변환되는 이유

the Numlevels parameter is used to scale the image into (even?) bins.
So when you would calculate the graycomatrix of a vector goint from 0 to 255 and you set Numlevels to 8. You should get 32 , 1's,2's,...,8's.

B(1,:) = 0:31;
B(2,:) = 32:63;
B(3,:) = 64:95;
B(4,:) = 96:127;
B(5,:) = 128:159;
B(6,:) = 160:191;
B(7,:) = 192:223 ;
B(8,:) = 224:255;
[m,SI] = graycomatrix(B,'NumLevels',8,'GrayLimits',[0 255]);
for i = 1 : 8
numel(find(SI==i))
end

INSTEAD : matlab returns following bins (as can be seen in SI)
Graylevel values
Bin1 = 0-18;
Bin2 = 19-45;
...
While they should all be 32 in size.

WHY ARE THEY NOT EVENLY SCALED ???

Sincerely

------------------------------

It's because of rounding - look at the SI matrix. Anything between 0
and 18 is less than 0.5 and so gets set to 1, anything from 19 to 45
gets scaled to between 0.5 and 1.5 which gets set to 2, and so on up
to 236 - 255 which gets scaled to 7.5 to 8 which gets set to 8.
You're assuming that the gray levels less than 32 when divided by 32
and multiplied by 8 all go to 1 and that is not true for the reasons
given above - only half of them do.

You will have only half as many numbers at the low end and the high
end because there are only half as many numbers that go into the
ranges 0-.5 and 7.5-8 as there are that go into [n-.5 to n+.5].