Copy columns from a matrix to another matrix in matlab - matlab

I have a matrix say ABC which has 60 rows and 120 columns.
Depending on another matrix X which is a 120 entry long array I would like to populate another matrix as follows:
if X(i)=1 column i is added to the matrix ABC_Copy.
if X(i)=0 column i is skipped the loop continues.
As it is obvious i would iterate from 1 to 120 which is the size of S representing the 120 columns in ABC.
How can we implement this in matlab without iterating entirely and placing each value individually?

You can use logical arrays for indexing in Matlab:
ABC_Copy = ABC(:, X==1);

Related

MATLAB matrix operation

I am having matrix with approx 3000 rows(changing) and 3 columns.
I have count of both rows and columns.
I am trying to plot the graph:
x=1:3000;
plot(matrix(x,1))
is there any way that I can include all rows in the plot instruction itself so that I can remove 'x=1:3000' ?
Also, I want to divide, 1st column of matrix which have 3000 rows into another matrix of 3 columns each with 1000 rows. Any specific instruction for this ?
I have made for loop for this and then i am placing individually the elements in the new array. But its taking long time.
As to the plotting issue, using the colon operator will plot all rows for your desired column:
plot(matrix(:,1));
EDIT: You mentioned you were a beginner. In case you haven't seen the colon operator used like this before, a colon operator all by itself when indexing into a matrix essentially means "all __", either "all rows" if in the first position or "all columns" if in the second position.
As for the second question, of splitting one column into a new matrix with multiple columns, you can use the reshape() function, which takes the input matrix to be reshaped and a number of output rows and columns. For example, to split the first column of matrix into 3 columns and put them into newMatrix, use the following:
newMatrix = reshape(matrix(:,1),[],3);
Note that the above code uses [] in the second argument (the number of rows argument) to mean "automatically determine number of rows".This is automatically determined based on the number of columns, which is defined in the third argument here as 3. The reshape function requires that the number of output rows * output columns be equal to input rows * input columns. So in the above case this will only work if the starting matrix has a number of rows which is divisible by 3.

Correlation and array in MATLAB

Kindly somebody help me in this.
I have two arrays of equal size 8x8.
And I need covariance of Column 1 of array 1 with the column 1 of second one.
After that I want to find Column 1 with the column 2 of second array.
After that I want to find Column 1 with the column 3 of second array.
After that I want to find Column 1 with the column 4 of second array.
After that I want to find Column 2 with the column 1 of second array.
And so on
I am assuming you want a measure as to how a column of first array varies with a column of second array. If yes, then that will be a scalar, otherwise, if you calculate the covariance matrix of two vectors, it is going to be a... matrix, obviously.
The following solution is based on the fact that covariance of two vectors is their correlation times the product of their standard deviations. More succinctly, for two random variables X and Y,
cov(X,Y)=corr(X,Y)*(sd(X)*sd*Y))
Thus, the solution to your question is:
pairCovariance=corr(X,Y).*(std(X).'*std(Y))

How to take the mean of columns from an array within an array?

I am currently working on a project in Matlab where I have a cell array of cell arrays. The first cell array is 464 columns long and 1 row deep. Each of these cells is another cell array that is 96 columns and 365 rows. I need to be able to get the mean of the 96 columns for each of the 464 arrays and place each of the 464 arrays on a different row in a new array called mean. I have tried to write code to just do one column as follow:
mean = Homes{1,1}(1:)
But I when ever I try to run this code I got the follow error:
mean = Homes{1,1}(1:)
|
Error: Unbalanced or unexpected parenthesis or bracket.
Basically my final array name mean needs to be 96 columns by 464 rows. I am stuck and could really use your help.
Thank you.
I suggest you to try the following code on a smaller matrix. See if it gives you the desired results.
a=cell(1,4); %for you it will be a=cell(1,464)
for i=1:4
a{i}=randi(10,[5 10]); %for you it will be a{i}=randi(10,[365 96]);
end
a1=cell2mat(a); %just concatenating
a1=mean(a1); %getting the mean for each column. in your case, you should get the mean for 96*464
a2=reshape(a1,[10 4]); %now what reshape does it it takes first 10 elements and arranges it into first column.
%Therefore, since I finally want a 4x10 matrix (in your case 464x96), i.e. mean of 10 elements in first cell array on first row and so on...
%Thus, 1st 10 elements should go to first column after doing reshape (since you want to keep them together). Therefore, instead of directly making matrix as 4x10, first make it as 10x4 and then take transpose (which is the next step).
a2=a2'; %thus you get a 4x10 matrix.
In your case specifically, the code will be
a=cell(1,464);
for i=1:464
a{i}=randi(10,[365 96]);
end
a1=cell2mat(a);
a1=mean(a1);
a2=reshape(a1,[96 365]);
a2=a2';

how can i deal with empty matrix: 0-by-any number resulting from simulation?

I have made a simulation and the result each time of the simulation is a matrix and i choose a certain row from the matrix, so if the simulation run=500, i'll have a 500 matrix and ,the row i choose each time will be (at the end of the simulation) 500 rows [one row from the first matrix...last row from the last matrix]...
the problem is some times a matrix dose not contain the certain row i want , the answer is for example empty matrix: 0-by-6
i want to ignor this answer
Note: the row i choose is not necessary to be exist in all matrices
so if the run=600 , result in 600 matrix , the row i choose maybe =400 only and the other 200 will be zero
the simulation STOP when the result is empty matrix: 0-by-any number
I use Matlab
you can use isempty to detect empty arrays, for example
a=zeros(0,5)
isempty(a)
a =
Empty matrix: 0-by-5
ans =
1
For when the index exceeds matrix dimensions, you can add a condition that tests the size of your matrix, specifically, how man rows using size(m,1)
So all together, in your for loop you can code something like:
for n=1:blah
if ~isempty(M) % continue if matrix is non-empty
if size(M,1)<=n % continue if index doesn't exceeds matrix dimensions
....
....

How to change elements in matrices using MATLAB

Starting wish a 7x4 binary matrix I need to change a random bit in each column to simulate error. Have been trying to no avail.
A very straightforward way to do this is to use a for loop. It might not be the most efficient approach in MATLAB, but it's probably good enough considering your data set is so small.
Iterate through each of the four columns. On each iteration, randomly chose a number from 1 to 7 to represent the row in that column that you have selected to change. Finally, flip the bit at that row/column. The following code does just this. Assume that "A" is a binary matrix with 7 rows and 4 columns
for col=1:4; %// Iterate through each column
row = ceil(7*rand()); %// Randomly chose a number from 1 to 7 to represent row
A(row,col) = ~A(row,col); %// Flip the bit at the specified row/col
end
Another possibility is to create 4 random numbers in one call, and assign in a vectorized fashion:
rowNumbers = randi(4,[1 4])
A(rowNumbers,:) = ~A(rowNumbers,:);