I have two arrays to plot, in x axis : array a. In y axis array b (ones and zeros)
I want to plot vertical lines when the values of array b =1
i.e
a=[23 12 76 43 21 90]
b=[1 0 1 1 0 1]
You may try bar plot.
a=[23 12 76 43 21 90];
b=[1 0 1 1 0 1];
bar(a,b)
Related
I have a matrix 2000x5, in the first column the point number, and in columns 2-5 the 4 neighbours (0s if there isnt a neighbour). Is there an efficient way to create an adjacency matrix out of this ?
1 129 0 65 0
2 130 0 66 85
3 131 169 67 0
4 132 170 68 87
5 133 0 69 81
6 134 0 70 82
7 135 173 71 83
8 136 174 72 84
9 137 161 73 0
10 138 162 74 93
11 139 163 75 0
12 140 164 76 95
13 141 165 77 89
14 142 166 78 90
15 143 167 79 91
16 144 168 80 92
17 145 0 81 65
18 146 0 82 66
....
I found the following thread, where it is explained for just one neighbour, but I am not sure how to use it for multiple neighbours.
matlab adjacency list to adjacency matrix
I would very much appreciate any help.
A quick and simple technique:
adjMat = zeros(size(A,1));
for ind = 1:size(A,1)
% Flag 1 on each row 'ind' at the indices mentioned in col 2-5
adjMat(ind, nonzeros(A(ind,2:end))) = 1;
end
Since you have mentioned using the nearest neighbour search, it is likely that the adjacency list should be completely filled to result in a undirected graph, in the sense that if row 1 has 20 as a neighbour, row 20 very likely has 1 as a neighbour.
However technically speaking, this will produce an adjacency matrix exactly equivalent to the adjacency list, assuming nothing by itself.
Example:
For an adjacency list
A = [1 2 3; 2 0 1; 3 1 4; 4 5 3; 5 4 0]
A =
1 2 3
2 0 1
3 1 4
4 5 3
5 4 0
The result is:
adjMat =
0 1 1 0 0
1 0 0 0 0
1 0 0 1 0
0 0 1 0 1
0 0 0 1 0
P.S. To force undirected-ness, you can simply add another statement in the for loop body:
adjMat(nonzeros(A(ind,2:end)),ind) = 1;
This will ensure that the adjacency matrix will be symmetric, which is a characteristic of undirected graphs.
Firstly, I'm going to assume that the adjacency list is undirected. In any case, it's not that far of a stretch to go to multiple neighbours. What you need to do first is detect the total number of non-zero elements per row from columns 2 to 5. Once you do this, for the rows of the adjacency matrix, you would copy the point number for as many times as there are non-zero elements per that row. The function repelem is perfectly suitable to do that for you. The column indices would simply be the second to fifth columns removing all of the zero elements. How you can do this is first transpose the matrix resulting in indexing the second to fifth columns, then using a logical indexing matrix to remove out the zero entries. Doing this will unroll your vector in a column-major fashion, which is why transposing is required before doing this operation. Once you do this, you can create row and column access indices so that these can be input into sparse much like that post you linked.
Supposing that your matrix was stored in A, you would do something like this. This also assumes that each of the weights connecting the nodes are 1:
% Find total number of non-zero elements per row, skipping first column
non_zero = sum(A(:,2:end) ~= 0, 2);
% Create row indices
rows = repelem(A(:,1), non_zero);
% Create column indices
cols = A(:,2:end).';
cols = cols(cols ~= 0);
% Create adjacency matrix
adj = sparse([rows; cols],[cols; rows], 1);
The above representation is in sparse. If you want the full numeric version, cast the output using full:
adj = full(adj);
If your graph is directed
If you have a directed graph instead of an undirected graph, the above call to sparse duplicates edges so that you are creating links to and from each of the neighbours. If your graph is actually directed, then you simply have to only use the row and column indices once instead of twice as seen in the above code:
% Create adjacency matrix
adj = sparse(rows, cols , 1);
Test Case
Here's a small test case to show you that this works. Supposing my adjacency list looked like the following:
>> A = [1 0 2 3; 2 4 0 0; 3 0 0 4]
A =
1 0 2 3
2 4 0 0
3 0 0 4
The adjacency matrix is now:
>> full(adj)
ans =
0 1 1 0
1 0 0 1
1 0 0 1
0 1 1 0
Taking a look at the list above and how the matrix is populated, we can verify that this is correct.
Note about repelem
repelem assumes you have MATLAB R2015a or later. If you don't have this, you can consult this answer by user Divakar on a custom implementation of repelem here: Repeat copies of array elements: Run-length decoding in MATLAB
I have a matrix A 4x10000, I want to use it to find another matrix C.
I'll simplify my problem with a simple example:
from a matrix A
20 4 4 74 20 20
36 1 1 11 36 36
77 1 1 15 77 77
3 4 2 6 7 8
I want, first, to find an intermediate entity B:
2 3 4 6 7 8
[20 36 77] 0 1 0 0 1 1 3
[4 1 1] 1 0 1 0 0 0 2
[74 11 15] 0 0 0 1 0 0 1
we put 1 if the corresponding value of the first line and the vector on the left, made a column in the matrix A.
the last column of the entity B is the sum of 1 of each line.
at the end I want a matrix C, consisting of vectors which are left in the entity B, but only if the sum of 1 is greater than or equal to 2.
for my example:
20 4
C = 36 1
77 1
N.B: for my problem, I use a matrix A 4x10000
See if this works for you -
%// We need to replace this as its not available in your old version of MATLAB:
%// [unqcols,~,col_match] = unique(A(1:end-1,:).','rows','stable') %//'
A1 = A(1:end-1,:).'; %//'
[unqmat_notinorder,row_ind,labels] = unique(A1,'rows');
[tmp_sortedval,ordered_ind] = sort(row_ind);
unqcols = unqmat_notinorder(ordered_ind,:);
[tmp_matches,col_match] = ismember(labels,ordered_ind);
%// OR use - "[tmp2,col_match] = ismember(A1,out,'rows');"
C = unqcols(sum(bsxfun(#eq,col_match,1:max(col_match)),1)>=2,:).'; %//'
%// OR use - "C = out(accumarray(col_match,ones(1,numel(col_match)))>=2,:).'"
This should work:
[a,~,c] = unique(A(1:end-1,:).', 'rows', 'stable');
C=a(histc(c,unique(c))>=2, :).';
Edit: For older versions of MATLAB:
D=A(1:end-1,:);
C=unique(D(:,squeeze(sum(all(bsxfun(#eq, D, permute(D, [1 3 2])))))>=2).', 'rows').':
I am new here and new to matlab.
I have a matrix file in matlab and what I want to do is make a plot of the average all the rows. However, I only want to plot a few data points (about 20) of the values before, and after, the maximum value within that row. The matrix file has 550 columns.
I have worked out how to identify the maximum value and the column number of this maximum value using;
[maxvalue maxindex] = max(filename, [], 2)
As the maximum is never in the same column, i really need help in calculating the average values for each row (before and afer the max value), and then how to plot this where the maximum value would be 0 on the x-axis.
For example - i have a matrix like this;
14 51 623 23 4 1 4 5
0 0 3 5 67 37 37 5
0 0 0 0 574 4 5 6
and max value = 623
67
574
and max index = 3
5
5
So i would like to, plot the average of the 3 rows, 2 data points before and after the max values...so to plot the average of;
14, 51, 623, 23, 4, 1
3, 5, 67, 37, 37
0, 0, 574, 4, 5
Thanks so much for any help!
data = [14 51 623 23 4 1 4 5
0 0 3 5 67 37 37 5
0 0 0 0 574 4 5 6]; %// example data
data = data.'; %'// it's easier to work along columns
[~, pos] = max(data); %// position of maxima
ind = bsxfun(#plus,bsxfun(#plus, pos,(-2:2).'),(0:size(data,2)-1)*size(data,1));
%'// linear index into the matrix obtained from pos
data_trimmed = data(ind).'; %'// index and transpose back
data = data.'; %'// undo transpose to put data back into shape
Result:
data_trimmed =
14 51 623 23 4
3 5 67 37 37
0 0 574 4 5
I need to simple plot B vs. A as bar plot in MATLAB, but I don't want my x axis showing completely from 1 to 274. I only need to show the existing data point on my x axis, which can be done easily in Excel as in the image below. How can MATLAB do this?
A=[1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 18 20 25 27 29 37 40 42 43 48 73 204 242 274];
B=[30 15 5 9 5 6 3 3 2 1 4 1 1 1 1 1 2 1 1 1 1 1 1 1 1 1 1 1 1 1];
You need to set both 'XTick' and 'XTickLabel' axes properties:
bar(B);
set(gca,'XTickLabel',A)
set(gca,'XTick',1:numel(A));
xlim([0 numel(A)+1]);
Here is an inelegant, but, nonetheless, working solution to your question:
x = [1,4, 6, 7]; % Your data
uni = unique(x)
yMax = length(find(x == mode(x))) + 1;
c = cell(1, length(uni));
c = strread(num2str(uni),'%s')
hist(1:length(uni));
axis([0 length(uni) 0 yMax])
set(gca, 'XTick', 1:length(uni));
set(gca, 'XTickLabel', c);
Basically, this plots the histogram as if the data were spread from 1 to the number of unique elements. Then, it sets the tick marks at each histogram value. Then, it labels each tick mark with the correct number.
I have some question. I have 2 matrix, it's have same size.
For example, first matrix :
1
1
0
0
1
0
Second matrix
34
56
12
12
33
14
Then, I want to compare this two matrix and groups it by the criteria on first matrix
so I will have this two groups matrix :
Matrix when the first matrix is have value 1
34
56
33
and
Matrix when the first matrix is have value 0
12
12
14
You could try this:
a = [1 1 0 0 1 0]';
b = [34 56 12 12 33 14]';
b(a==1)
b(a==0)