How to implement code based on the following algorithm?

1. Assign the data (90 projects) to be used in developing the MLPmodel.
2. The remaining 70 projects will be used to evaluate the MLPmodel (Section 4).
3. Randomly divide the 90 data points into 10 equal sets (S1 toS10).
4. Set the number of nodes in the hidden layer to 10 (“nh” = 10).
5. Set the number of training rounds (i) to 1 (“i” = 1)
6. In Round “i” (“i” is a number between 1 and 10), use 9 setsfor training and validation and 1 set for testing (for each valueof “i”, 9 different sets are used for training/validation and theremaining set for testing)
7. Record the testing error Vi-nh (“i” represents the number ofthe round, and “nh” the number of nodes in the hidden layer.For instance, the first testing error will be V1-10).
8. Increment the value of “i” by 1.
9. If the value of “i” is 11, then increment the value of “nh” by 1and set the value of “i” to 1.
10. If the value of “nh” = 19, then stop training process and exit.
11. Go to step “6”

[Image Analyst]

http://www.mathworks.com/matlabcentral/answers/8026-best-way-s-to-master-matlab

You might look at datasample() or randsample().