-3

Can someone help me. What I need to do if I want to convert a struct array to double or complex double to double because i need to use it as input for svm classifier.Thanks in advanced .

Aida Ezati
  • 21
  • 7
  • Please show us your code so far... – Paulo Mattos Apr 19 '17 at 03:11
  • i already combined 7 patients that contain 19 feature extraction each one.. the data in 7x1 struct with 19 field ..when i using in SVM clc;clear all;close all; load trainset.mat data =new_var; group = label; SVMStruct = svmtrain(data,group,'kernel_function','linear'); species = svmclassify(SVMStruct,meas,'showplot',false); the error will be Error using svmtrain (line 241) TRAINING must be a numeric matrix. i know that i need to convert the struct array into double but dont know how? can u help me – Aida Ezati Apr 19 '17 at 03:23
  • 1
    It's much more helpful if you add your code to the initial question with proper formatting. Reading it out of the comments is...difficult. It would probably also help to see a small extract of the structure that's causing the error. – user2027202827 Apr 19 '17 at 07:35

1 Answers1

0

Assuming your data struct looks like this (with 7 patients, 19 features and 150 feature vectors for each patient):

data = 

    p1: [150x19 double]
    p2: [150x19 double]
    p3: [150x19 double]
    p4: [150x19 double]
    p5: [150x19 double]
    p6: [150x19 double]
    p7: [150x19 double]

and you want to concatenate them into one big matrix (you will not distinguish between patients when training your SVM) of the form 1050x19 double, then the following code snippet will do the trick:

% Generate test data
data.p1(:,1:19) = randn(150,19);
data.p2(:,1:19) = randn(150,19);
data.p3(:,1:19) = randn(150,19);
data.p4(:,1:19) = randn(150,19);
data.p5(:,1:19) = randn(150,19);
data.p6(:,1:19) = randn(150,19);
data.p7(:,1:19) = randn(150,19);

% Add each patient features to cell array
data_cell = cellfun(@(field) data.(field), fieldnames(data), 'UniformOutput', false);

% Vertically concatenate the entries in the cell array from above 
data_combined = vertcat(data_cell{:}); %1050x19 double

You will need to concatenate your group labels in a similar way as well, making sure you don't loose track of which group label corresponds to which feature vector in your combined data.

Best of luck

Emil Østergård Johansen
  • 31
  • 6
  • thanks but what is feature vector? I have 7x1 struct with 19 field , which one i should assume as feature vector? – Aida Ezati Apr 20 '17 at 01:35