what is the value of predicted and actual?

function EVAL = Evaluate(ACTUAL,PREDICTED)
% This fucntion evaluates the performance of a classification model by 
% calculating the common performance measures: Accuracy, Sensitivity, 
% Specificity, Precision, Recall, F-Measure, G-mean.
% Input: ACTUAL = Column matrix with actual class labels of the training
%                 examples
%        PREDICTED = Column matrix with predicted class labels by the
%                    classification model
% Output: EVAL = Row matrix with all the performance measures

idx = (ACTUAL()==1);

p = length(ACTUAL(idx));
n = length(ACTUAL(~idx));
N = p+n;

tp = sum(ACTUAL(idx)==PREDICTED(idx));
tn = sum(ACTUAL(~idx)==PREDICTED(~idx));
fp = n-tn;
fn = p-tp;

tp_rate = tp/p;
tn_rate = tn/n;

accuracy = (tp+tn)/N;
sensitivity = tp_rate;
specificity = tn_rate;
precision = tp/(tp+fp);
recall = sensitivity;
f_measure = 2*((precision*recall)/(precision + recall));
gmean = sqrt(tp_rate*tn_rate);

EVAL = [accuracy sensitivity specificity precision recall f_measure gmean];

[Monisha R]

What was the right code for this?

Answer 1

[Star Strider]

From the documenbtation:

% Input: ACTUAL = Column matrix with actual class labels of the training
%                 examples
%        PREDICTED = Column matrix with predicted class labels by the
%                    classification model

I would assume that ‘matrix’ is ‘vector’ here. The ‘ACTUAL’ values are whatever the original classifications are, and the ‘PREDICTED’ the results of your classifier. The values likely depend on the original ‘correct’ classifications and your classifier output.

[oza san]

dear Star i have been trying to use the above code but i stack with an error "index exceeds matrix dimensions. Error in Evaluate1 (line 21) tp = sum(ACTUAL(idx)==d(idx));" what can i do?

if true
 newX=load('C:\Users\sam\Desktop\dataset\pred.mat');%to load the classifier output!
d = struct2cell(newX); % to change its format from struct to cell array
ACTUAL = zeros(2000,1); the orginal value
ACTUAL(1:1000) = 0;% setfirst 1000 row as negative sapmle
ACTUAL(1001:end) = 1; % set the rest as postive sample
the rest of the code is as it is!!
end

[Star Strider]

You must be certain your ‘ACTUAL’ and ‘d’ vectors are the same sizes (same number of rows and columns in each).

[oza san]

Dear Star thank you for your fast replay.ok.i will try that way but since ACTUAL is the value of the training sample how it could be equal with testing sample which is 20% of the training sample?

[Star Strider]

Use the testing sample and the classifications from it for both vectors.

[oza san]

ya i test it manually and works fine.before i give the the test sample for the classifier i set the values for it[0 and 1]. then i get the classifier output with the same column number finally i use both of them for evaluation.

[Star Strider]

If this helped you solve your problem, please Vote for my Answer!

[oza san]

sure i did it sir!!!thank you so much!!!

[Star Strider]

Thank you!

As always, my pleasure!

[oza san]

Dear sir,here i am again i need to extract a word from a sentence in an image using vertical projection profile mind you if you have some ample codes?

[Star Strider]

Post this as a new Question.

Image processing is not an area of my expertise.

I cannot help you with it.

[oza san]

ok thank you!!i will search!!!

[Star Strider]

As always, my pleasure!

[oza san]

Dear Sir, i tried to load my actual and predicted values to calculate performance measures but i can't load the data.i get an error"Undefined operator '==' for input arguments of type 'cell'."!! actually when i give the actual manually it woks fine!!!the problem is it can not load the data found in .mat file. what can i do?

if true
  % code
ACTUAL=load('C:\Users\sam\Desktop\dataset\ACTUAL.mat');%load actual
pred=load('C:\Users\sam\Desktop\dataset\pred.mat');%load predicted
c = struct2cell(ACTUAL);%change the struct to cell
d = struct2cell(pred);%change the struct to cell
idx = (c()==1);
p = length(c(idx));
n = length(c(~idx));
N = p+n;
tp = sum(c(idx)==d(idx));
tn = sum(c(~idx)==d(~idx));
fp = n-tn;
end

[Star Strider]

The only place I see that is in your ‘idx’ assignment.

This should work:

c = {1 2 6 1 3 7 9 1};                                                      % Create Test Data
idx = cellfun(@eq, c, mat2cell(ones(size(c)), 1, ones(size(c))), 'Uni',0);
idx = [idx{:}];

[oza san]

i cam up with an error"Input arguments, D1 through D2, must sum to each dimension of the input matrix size, [200 1].'"

if true
  % code
c= zeros(200,1); creating test data
c(1:end) = 1;
idx = cellfun(@eq, c, mat2cell(ones(size(c)), 1, ones(size(c))), 'Uni',0);
idx = [idx{:}]

end

[oza san]

hello sir do i have to post as new question so that you can get a vote?

[Star Strider]

You didn’t tell me what size ‘c’ is.

Try this:

idx = cellfun(@eq, c, mat2cell(ones(size(c)), ones(size(c)), 1), 'Uni',0);

[oza san]

still not working i get an error"input #2 expected to be a cell array, was double instead"

[Star Strider]

This works correctly for me:

c = {1 2 6 1 3 7 9 1}';                                                      % Create Test Data
idx = cellfun(@eq, c, mat2cell(ones(size(c)), ones(size(c)), 1), 'Uni',0)
idx = [idx{:}]

I cannot determine what the problem may be with respect to your code.

I am using R2017a. There could be version differences in the way the functions work.

[oza san]

ok.i am using R2016a!!! i change the value of c to cell using this myCell={c};but i face Index exceeds matrix dimensions.even though i check it.

[Star Strider]

I thought ‘c’ was already a cell array. That was the original problem.

What is the result when you run this from your Command Window (or script):

whos c

[oza san]

here is what i get Name Size Bytes Class Attributes

c         200x1              1600  double              

Name        Size            Bytes  Class    Attributes

d         200x1             22800  cell

[oza san]

i think c is double both before and after conversion attempt!!!

[Star Strider]

I don’t understand that in the context of these earlier assignments:

c = struct2cell(ACTUAL);%change the struct to cell
d = struct2cell(pred);%change the struct to cell

[oza san]

i thought that both actual and predicted value are structure so i tried to convert to cell array.but now i ignore.

[Star Strider]

I’m lost.

If ‘c’ is a double array, your original:

idx = c == 1;

should work.

[oza san]

 " Undefined operator '==' for input arguments of type 'cell'."
tp = sum(c(idx)  == d(idx));what is wrong with this one?

[oza san]

Dear sir i make you too busy let me see it by myself and if i can't i will be back.

[oza san]

Dear sir.here i am to say thank you!!!just i change the cell array to double matrix using "d = str2double(d);" and it works fine.even though am not sure other commands(performance measurements) will not be affected.

[Star Strider]

My pleasure.

[Nowshin Jenny]

Dear sir, after using this command in my code for EEG signal classification, I got following error-

"Function definitions are not permitted in this context."

How can i solve it?