MATLAB and Simulink resources for Arduino, LEGO, and Raspberry Pi

# NCHOOSEK - What am I doing wrong?

Asked by Simon on 24 May 2013
```V = nchoosek({CRY,DXY,GOLDS,USCRWTIC,USGG2YR,USGG10YR},4);
```
```for i= 1:15;
```
`      A{i} = V(i,:);`
```end;
```

Basically, I would need to create A1 to A15 containing the 'i' line of the matrix V. For example:

```V = [1 2 3 4 ; 5 6 7 8 ; 9 0 1 2]
```

I would need

```A1 = [1 2 3 4]
A2 = [5 6 7 8]
A3 = [9 0 1 2]
```

I've been searching for this for all day >.<

## Products

No products are associated with this question.

Answer by Image Analyst on 25 May 2013

Sure, if you want to regress, just run this demo (where I made up some arbitrary values):

```CRY = rand(1, 20);
DXY = rand(1, 20);
GOLDS = rand(1, 20);
USCRWTIC = rand(1, 20);
USGG2YR = rand(1, 20);
USGG10YR = rand(1, 20);
```
```V = nchoosek({CRY,DXY,GOLDS,USCRWTIC,USGG2YR,USGG10YR},4)
```
```for i= 1:15
A{i} = V(i,:);
end
y = sin(1:20);
% Regress y against A{1}
% Get one cell array of 4 cells from the first element of A
caRowOne = A{1}
% Extract just the first of those 4 cells.
xTraining = caRowOne{1}
% Plot training data.
plot(xTraining, y, 'bo');
hold on;
% Do a fit to a line.
coefficients = polyfit(xTraining, y, 1);
% See how well we did.
x = linspace(min(xTraining(:)), max(xTraining(:)), 40);
fittedY = polyval(coefficients, x);
plot(x, fittedY, 'r-', 'LineWidth', 2);
grid on;
title('Y vs. A{1}', 'FontSize', fontSize);
xlabel('X', 'FontSize', fontSize);
ylabel('Y', 'FontSize', fontSize);
```

## 1 Comment

Simon on 26 May 2013

Pretty cool stuff Brah! I will try to loop this for all 15 combinations and then a p-value, F and t-test for each. Thanks a lot!

Answer by Azzi Abdelmalek on 24 May 2013
Edited by Azzi Abdelmalek on 24 May 2013
```V = [1 2 3 4 5 6 7 8 9 0 1 2];
V=reshape(V,4,[]);
for i= 1:3
A{i} = V(:,i)';
end
```

Azzi Abdelmalek on 24 May 2013

Why do you want to create A1,A2 and A3, while one variable A is enough ?

Azzi Abdelmalek on 24 May 2013

If you insist to create A1,A2 and A3 (which is not recomanded)

```V = [1 2 3 4 5 6 7 8 9 0 1 2];
V=reshape(V,4,[]);
for i= 1:3
eval([ sprintf('A%d',i) '= V(:,i)'''])
end
```
Jan Simon on 25 May 2013

@Simon: Don't use EVAL. Creating A1, A2, ... is a bad programming pattern and therefore the most cited question from the FAQ: FAQ: How can I create A1, A2, ... in a loop .

Staying with the original matrix and using V(i, :) would be much more direct and efficient instead of inserting indirect intermediate copies to a bunch of variables, which require crude tricks to be access automatically.

Answer by Andrei Bobrov on 24 May 2013
```A = num2cell(V,2);
```

Answer by Jan Simon on 25 May 2013

V(i, :) is perfect already. Are you sure that there is any reason for adding a level of abstraction by splitting the matrix to a bunch of vectors?

Simon on 25 May 2013

Will i be able to perform regression on each cell of A(i) with another variable? (cell 1,1 with Y , then Cell 1,2 with Y , etc)

Jan Simon on 25 May 2013

@Simon: It is very likely, that you can perform the regression even with the original matrix directly. The most Matlab commands can handle a matrix input and process the operations for the subvectors automatically. When you post the code you want to apply, more detailed suggestions are possible.

When you want to access the contents of a cell, use the curly braces: A{1} etc.

Answer by Simon on 26 May 2013

Thanks everyone! If could accept everyone's answer i would..

Randy Souza on 31 May 2013

Simon, you can always vote for the other answers!

Simon on 31 May 2013

How?

Randy Souza on 31 May 2013

There should be a triangle with "x Votes" under the profile picture for the answerer. Click that and it should change to "(x + 1) Votes"