How to make code to split, compute mean, apply Softmax

I have 90 datasets (9 data x 10 labels)

1. split the dataset into support(80dataset) and query(10dataset)

2. Compute each mean of examples(9 means)

3. Compute the Euclidean distance between each mean and query(10dataset)

4. apply Softmax and calculate probabilities

5. compute accuracy

[Sindar]

check out splitapply: https://www.mathworks.com/help/matlab/ref/splitapply.html

Answer 1

[Gaurav Garg]

Hi,

You can use splitapply function to split your whole data into 9 groups and apply the mean function to each group. It would return you an array of 9 elements, where each element is a mean to one group. You can now, carry on with the third step to compute the Euclidean distance between each mean and the query set, and proceed with steps 4 and 5.

Algo:

Y=splitapply (mean, X, G); % G is a vector of group numbers, X is the whole data
for i=1:8
         % compute Euclidean distance between Y[i] and query dataset
end
% Steps 4 and 5

[Kong]

Thank you so much.

Could you let me know how to find Group (G)?

I have a dataset (90 x 2857), The last column (2857) is class.

[Sindar]

If your dataset is X and the last column is not part of the data:

G = findgroups(X(:,end));
Y = splitapply(mean, X(:,1:end-1), G);

[Gaurav Garg]

Hi Kong,

Kindly go through the link here.