What is difference between training set and test set database?

I have collection of database having images with 25 categories like mountains,rivers,fields,chaparal ,etc .Each category have 40 images.Now I dont know how to divide the dataset into training and test set.In my project a query image from query folder is fired and from the dataset the similar images will be given as output.

Pls suggest how to divide the dataset in good ratio.

Answer 1

[Walter Roberson]

http://uk.mathworks.com/matlabcentral/answers/216336#answer_178776

http://uk.mathworks.com/matlabcentral/answers/217182-divide-data-set-in-matlab-code

[preeti]

Thank you Sir,

But I have dataset of 1000 images with 25 categories

I get this code

nrows = size(YourData,1);

r80 = round(0.80 * nrows);
rand80 = randperm(nrows,r80);

trainingset = YourData(rand80,:);

testset = YourData;

testset(rand80,:) = [];

but wat should be urdata?

[Walter Roberson]

Suppose that you read the images into I{K} for the K'th image, so I is a cell array of data arrays. And suppose that all of the images are exactly the same size. Then

T = cellfun(@(img) img(:), I);  %make columns from each image
T2 = horzcat(T{:});  %splice the columns together
YourData = T2.';  %so each image is now a row vector