how to specify the input and target data

I have a dataset 2310x25 table. I dont know how to specify the input and target data. i'm using the below code for k fold cross validation.

data= dlmread('data\\inputs1.txt'); %inputs
groups=dlmread('data\\targets1.txt'); % target
Fold=10;
indices = crossvalind('Kfold',length(groups),Fold);
for i =1:Fold
    testy = (indices == i);   
    trainy = (~testy);   
    TestInputData=data(testy,:)'; 
    TrainInputData=data(trainy,:)';
    TestOutputData=groups(testy,:)'; 
    TrainOutputData=groups(trainy,:)';

[Walter Roberson]

dlmread() always returns a numeric array, never a table() object.

See the example https://www.mathworks.com/help/bioinfo/ref/crossvalind.html#mw_54071adb-b4b0-462b-a1c3-8cc32d7085d0

[uma]

my question is not about dlmread function, i can use there xlsread(segment.csv). the problem is how to specify the input and target data.

[Walter Roberson]

that example shows

    test = (indices == i); 
    train = ~test;
    class = classify(meas(test,:),meas(train,:),species(train,:));

This assumes numeric arrays. The code would have to be modified if the input is a table like you posted. We would need to know which table variables stored the information of interest.

[uma]

All table varaibles are important, there are 7 classes in the table that i want to use for classification purpose.

[Walter Roberson]

Are you working with a table() object or with something read by xlsread? Are all of the columns numeric? Where is the information about the class stored?

[uma]

I have attached my dataset you can check it .I used it to test the performance of my model.

[Walter Roberson]

Are you aware that some of the entries are question mark?

[uma]

yes I know that. Now can you tell me how this dataset can be used to specify the input and target data

Answer 1

[Walter Roberson]

filename = 'https://www.mathworks.com/matlabcentral/answers/uploaded_files/1038775/bankruptcy.csv';
opt = detectImportOptions(filename, 'TrimNonNumeric', true);
data = readmatrix(filename, opt);
data = rmmissing(data);

groups = data(:,end);
data = data(:,1:end-1);

whos groups

  Name           Size            Bytes  Class     Attributes

  groups      3194x1             25552  double              

[sum(groups==0), sum(groups==1)]

ans = 1×2

3164 30

cp = classperf(groups);

Fold=10;
indices = crossvalind('Kfold',length(groups),Fold);
failures = 0;

for i =1:Fold
    test = (indices == i); 
    train = ~test;
    try
        class = classify(data(test,:), data(train,:), groups(train,:));
        classperf(cp, lass, test);
    catch ME
        failures = failures + 1;
        if failures <= 5
            fprintf('failed on iteration %d\n', i);
        else
            break
        end
    end
end

failed on iteration 1
failed on iteration 2
failed on iteration 3
failed on iteration 4
failed on iteration 5

cp

                        Label: ''
                  Description: ''
                  ClassLabels: [2×1 double]
                  GroundTruth: [3194×1 double]
         NumberOfObservations: 3194
               ControlClasses: 2
                TargetClasses: 1
            ValidationCounter: 0
           SampleDistribution: [3194×1 double]
            ErrorDistribution: [3194×1 double]
    SampleDistributionByClass: [2×1 double]
     ErrorDistributionByClass: [2×1 double]
               CountingMatrix: [3×2 double]
                  CorrectRate: NaN
                    ErrorRate: NaN
              LastCorrectRate: 0
                LastErrorRate: 0
             InconclusiveRate: NaN
               ClassifiedRate: NaN
                  Sensitivity: NaN
                  Specificity: NaN
      PositivePredictiveValue: NaN
      NegativePredictiveValue: NaN
           PositiveLikelihood: NaN
           NegativeLikelihood: NaN
                   Prevalence: NaN
              DiagnosticTable: [2×2 double]

[Walter Roberson]

The reason for the failure is that you only have 30 entries with class 1, and when you are doing random selection for K-fold purposes, you are ending up with situations where there are no entries for class 1 in the training data.