i have trained a classifier with a data, but now i dont know how to test it for other data.

main function--

DataSet = readtable('Dataset.xlsx');

%DataSet bir Tablo nesnesidir

%iknci parametre fold parametresi

fold=342;

N=5;

D_orani=zeros(N,3);

k1=zeros(2,2,N);

k2=zeros(2,2,N);

k3=zeros(2,2,N);

for i=1:N

[k1(:,:,i),D_orani(i,1)]=Ensemble_GentleBoost(DataSet,fold);

[k2(:,:,i),D_orani(i,2)]=KnnClassifier(DataSet,fold);

[k3(:,:,i),D_orani(i,3)]=LojisticRegression(DataSet,fold);

end

Dogruluk_orani=sum(D_orani,1)/N;

k1=mean(k1,3);k2=mean(k2,3);k3=mean(k3,3);

fprintf('Ensemble_GentleBoost icin= %8.8f ',Dogruluk_orani(1)*100)

fprintf('\n');

fprintf('Weighted K-NN= %8.8f %',Dogruluk_orani(2)*100)

fprintf('\n');

fprintf('Lojistic Regression icin= %8.8f ',Dogruluk_orani(3)*100)

fprintf('\n');

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

knnclassifier.m

%*** 24/12/2018*********************************************%

%*** ALHASAN ALKHATIB B140100255****************************%

%*** Ses kayitlari ile Parkinson hastaligi tespiti**********%

%*** KnnClassifier.m dosyasi********************************%

%***********************************************************%

% Weighted K-NN

function [konfizyon,validationAccuracy]=KnnClassifier(DataSet,fold)

inputTable = DataSet;

predictorNames = {'Jitta', 'jitt', 'jit_rap', 'jit_ppq5', 'jit_DDP', 'sh_DB', 'shimmer', 'sh_apq3', 'sh_apq5', 'sh_apq11', 'shim_DDP', 'median_pitch', 'mean_pitch', 'max_pitch', 'min_pitch', 'range_pitch', 'variation', 'oto_KT', 'oto_K0', 'HNR', 'NHR'};

predictors = inputTable(:, predictorNames);

response = inputTable.class;

% Train a classifier

classificationKNN = fitcknn(...

predictors, ...

response, ...

'Distance', 'Euclidean', ...

'Exponent', [], ...

'NumNeighbors', 11, ...

'DistanceWeight', 'SquaredInverse', ...

'Standardize', true, ...

'ClassNames', [0; 1]);

% predict function

predictorExtractionFcn = @(t) t(:, predictorNames);

knnPredictFcn = @(x) predict(classificationKNN, x);

trainedClassifier.predictFcn = @(x) knnPredictFcn(predictorExtractionFcn(x));

trainedClassifier.RequiredVariables = {'Jitta', 'jitt', 'jit_rap', 'jit_ppq5', 'jit_DDP', 'sh_DB', 'shimmer', 'sh_apq3', 'sh_apq5', 'sh_apq11', 'shim_DDP', 'median_pitch', 'mean_pitch', 'max_pitch', 'min_pitch', 'range_pitch', 'variation', 'oto_KT', 'oto_K0', 'HNR', 'NHR'};

trainedClassifier.ClassificationKNN = classificationKNN;

trainedClassifier.About = 'This struct is a trained classifier exported from Classification Learner R2016a.';

trainedClassifier.HowToPredict = sprintf('To make predictions on a new table, T, use: \n yfit = c.predictFcn(T) \nreplacing ''c'' with the name of the variable that is this struct, e.g. ''trainedClassifier''. \n \nThe table, T, must contain the variables returned by: \n c.RequiredVariables \nVariable formats (e.g. matrix/vector, datatype) must match the original training data. \nAdditional variables are ignored. \n \nFor more information, see <a href="matlab:helpview(fullfile(docroot, ''stats'', ''stats.map''), ''appclassification_exportmodeltoworkspace'')">How to predict using an exported model</a>.');

% predictors and response

inputTable = DataSet;

predictorNames = {'Jitta', 'jitt', 'jit_rap', 'jit_ppq5', 'jit_DDP', 'sh_DB', 'shimmer', 'sh_apq3', 'sh_apq5', 'sh_apq11', 'shim_DDP', 'median_pitch', 'mean_pitch', 'max_pitch', 'min_pitch', 'range_pitch', 'variation', 'oto_KT', 'oto_K0', 'HNR', 'NHR'};

predictors = inputTable(:, predictorNames);

response = inputTable.class;

% cross-validation

partitionedModel = crossval(trainedClassifier.ClassificationKNN, 'KFold', fold);

% validation accuracy

validationAccuracy = 1 - kfoldLoss(partitionedModel, 'LossFun', 'ClassifError');

% validation predictions and scores

[validationPredictions, validationScores] = kfoldPredict(partitionedModel);

[N,F]=size(DataSet);

R=response;

VP=validationPredictions;

AA=0;AN=0;NA=0;NN=0;

for i=1:N

if R(i)==1 && VP(i)==1

AA=AA+1;

elseif R(i)==1 && VP(i)==0

AN=AN+1;

elseif R(i)==0 && VP(i)==1

NA=NA+1;

elseif R(i)==0 && VP(i)==0

NN=NN+1;

end

end

konfizyon=zeros(2,2);

konfizyon(1,1)=AA/168;konfizyon(1,2)=AN/168;konfizyon(2,1)=NA/174;konfizyon(2,2)=NN/174;

end

Answer 1

[TED MOSBY]

Hi Gaurav,

I see that you have trained a classifier using some data and now you seek a function to test out the model on a new set of data. To do so, you can leverage the "predict" function which can be called as follows:

%............your above code…..

predictions = predict(your_trained_model_name, newData);

disp(predictions);

%.....end…..

Your “newData” should be in the same format and should have the same number of features as your training data.

That should work, alternatively you can try appending your new data in the older data and then split the data for training and test accordingly. Below is an example code on how you can split the data into training and testing data:

% Append new test data to the training data 
X = [X; newTestData]; 
Y = [Y; cell(size(newTestData, 1), 1)];  % Assign placeholder labels for the new test data 

% Specify the indices for the training set 
trainIndices = 1:size(X, 1) - size(newTestData, 1); 

% Specify the indices for the testing set 
testIndices = size(X, 1) - size(newTestData, 1) + 1:size(X, 1); 

% Training set 
XTrain = X(trainIndices, :); 
YTrain = Y(trainIndices, :); 

% Testing set 
XTest = X(testIndices, :); 
YTest = Y(testIndices, :); 

% Train a k-Nearest Neighbors classifier 
knnModel = fitcknn(XTrain, YTrain, 'NumNeighbors', 3); 

% Predict using the trained classifier on the test set 
predictions = predict(knnModel, XTest); 

% Display the predictions 
disp('Predictions for test data:'); 
disp(predictions); 

I hope it helps. You may refer to the MATLAB R2018a documentation for the predict function here:

https://in.mathworks.com/help/releases/R2018b/stats/classificationknn.predict.html

You can also refer this for more information on splitting the data:

https://in.mathworks.com/help/releases/R2018b/stats/surrogate-splits-for-missing-data.html?searchHighlight=XTest&s_tid=doc_srchtitle