how to cluster bit strings using matlab

A= [ 1 0 0 0 0 0 1 1 0,
     0 0 0 0 1 1 0 0 1,
     1 1 1 1 1 1 1 0 0,
     1 1 1 1 1 1 1 0 0,
     1 0 0 0 0 0 1 1 0,
     0 0 0 0 1 1 0 0 1,
     1 1 1 1 1 1 1 1 1]

let A be the binary matrix.. 1 means there is decrease in production and 0 means there is increase in production ..

how will i cluster the same bit strings patterns?

i m working on health care data set[rows= 453 and col = 60] .. 1 represent decrease and 0 represent increase ..can this be possible using loops..

note: order of bits are important..// order dependent

[KSSV]

what do you mean by cluster? You want the positions of 1 and 0?

[kumud alok]

i mean similar bit strings should be in one cluster or group or it should be saved in another variable.

[KSSV]

you can extract 0 and 1 from any matrix using find(mymatrix==1). It will give positions also. Then you can save..

[kumud alok]

sir, this function find(mymatrix==1) gives only the location of 1...i got the location ..but what i want is..from matrix A..there are 7 bit strings separated by commas. bit strings 3 and 4 should be one group because pattern are similar ..bit string 1 and 5 shows similar pattern therefore it should be in one cluster(cluster means group of similar things)..

[kumud alok]

can u suggest me any clustering algorithm ..

[Guillaume]

A = [ 1 0 0 0 0 0 1 1 0, 0 0 0 0 1 1 0 0 1, 1 1 1 1 1 1 1 0 0, 1 1 1 1 1 1 1 0 0, 1 0 0 0 0 0 1 1 0, 0 0 0 0 1 1 0 0 1, 1 1 1 1 1 1 1 1 1]

"there are 7 bit strings separated by commas". No, there isn't. Please learn the matlab syntax, a comma is exactly the same as a space. A is a 1 x 63 vector.

If you were to use a semicolon between each group of 9, then A would be a 7 x 9 matrix:

A = [ 1 0 0 0 0 0 1 1 0; 0 0 0 0 1 1 0 0 1; 1 1 1 1 1 1 1 0 0; 1 1 1 1 1 1 1 0 0; 1 0 0 0 0 0 1 1 0; 0 0 0 0 1 1 0 0 1; 1 1 1 1 1 1 1 1 1]

Or you could put your 7 bit strings into a 1 x 7cell array containing 1 x 9 vectors

A = {[ 1 0 0 0 0 0 1 1 0], [0 0 0 0 1 1 0 0 1], [1 1 1 1 1 1 1 0 0], [1 1 1 1 1 1 1 0 0], [1 0 0 0 0 0 1 1 0], [0 0 0 0 1 1 0 0 1], [1 1 1 1 1 1 1 1 1]}

[kumud alok]

okay..thank you so much sir...i m learning matlab..i m working in a project ..my work is to analyse bit string patterns and then apply clustering algorithm..i have applied hierarchical clustering but i m still confused..i just want to asked u about the clustering ..and distance measure also..which similarity measure used i should either hamming or lavenstein..please help me..

[Walter Roberson]

The original code used

A= [ 1 0 0 0 0 0 1 1 0,
     0 0 0 0 1 1 0 0 1,
     1 1 1 1 1 1 1 0 0,

and so on. MATLAB does know to treat the end of line line a semi-colon in that case.

Answer 1

[Guillaume]

It sounds like you just want to identify all identical 'bit strings' and assign them the same ID. This can easily be achieved with unique.

First, let's change A into something useful, since as mentioned in the comment to your question A is not "7 bit strings separated by commas", matlab ignores the commas. While we're at it, let's give the variable a name that has meaning:

A = [ 1 0 0 0 0 0 1 1 0, 0 0 0 0 1 1 0 0 1, 1 1 1 1 1 1 1 0 0, 1 1 1 1 1 1 1 0 0, 1 0 0 0 0 0 1 1 0, 0 0 0 0 1 1 0 0 1, 1 1 1 1 1 1 1 1 1]
bitstreams = reshape(A, 9, [])'; %reshape A into an n x 9 array

To give a unique ID to each identical bit stream we can simply use the third return value of unique, with the 'rows' option to the unique to consider each row as a complete object.

[~, ~, id] = unique(bitstreams, 'rows', 'stable') %stable optional

This returns

id = [1; 2; 3; 3; 1; 2; 4]

where you can see that the 1st and 5th bitstream have been assigned id 1, the 2nd and 6th id 2, the 3rd and 4th id 3, and the last one id 4.

[kumud alok]

thank u so much sir...