Filter löschen
Filter löschen

Replace value in matrix

1 Ansicht (letzte 30 Tage)
Nicolas
Nicolas am 14 Mär. 2011
Hi,
I'm looking at the orientation of lines generating using linear regression.
I create a matrix that increases after each loop, generating a new columns with a number of data higher than the previous columns (because there is more lines).
My problem is that in the end my matrix is [24*17], however the data in the first colum are only 3 values, and the rest (19 cells) filled with zeros.
I'd like to analyse the matrix, but i can'tget rid of the "filling-zeros".
Does anyone have an idea?
Thanks
N.
  3 Kommentare
Walter Roberson
Walter Roberson am 14 Mär. 2011
Is the matrix the _result_ of linear regression, or is it input being fed into linear regression ? If it is input to linear regression, what would you intend it to mean to the formula -- that the corresponding components are zero ?
Nicolas
Nicolas am 16 Mär. 2011
it is the results of linear regression.

Melden Sie sich an, um zu kommentieren.

Akzeptierte Antwort

Oleg Komarov
Oleg Komarov am 16 Mär. 2011
You can obtain:
A = [0 0 143 143 152
0 0 0 0 151
0 0 0 0 143];
regexprep(evalc('A'), '0', ' ')
A =
143 143 152
151
143
But it's a non tractable string array, i.e. what you're trying to do is meaningful only for visualization purposes and privates A of any computational use.
Oleg
  5 Kommentare
Nicolas
Nicolas am 16 Mär. 2011
regexprep(A, '0', ' ')
Nicolas
Nicolas am 16 Mär. 2011
Sorry, my explanations are a bit confusing.
My first idea was to analyse the matrix, by "analysing" i meant plotting the direction values of each column in a compass plot (all the data, then by sector of 10 degrees) to see preferential orientations. The problem that you helped me to solve was to get rid of the "filling-zeros", because my matrix is growing in each loop both row and column.
I hope it is a bit more clear for you.

Melden Sie sich an, um zu kommentieren.

Weitere Antworten (2)

Paulo Silva
Paulo Silva am 14 Mär. 2011
Two options:
1-ignore first column
mat(:,2:end)
2-replace the zeros on the first column
mat(mat(:,1)==0)=inf %replace zeros with inf
you can choose other things besides inf, maybe NaN or a really small value like eps
3-extra option, replace all zeros
mat(mat==0)=inf %again you can choose the value to replace the zeros

Nicolas
Nicolas am 16 Mär. 2011
Hi,
Thank you for your help. It works very well!
this is my matrix
0 0 143 143 152
0 0 0 0 151
0 0 0 0 143
I was trying to use indices.
[r,c,v]=(mat>0)
which gives me
1 3 143
1 4 143
1 5 152
2 5 151
3 5 143
can i rebuilt a matrix like the following using the indices
143 143 152
151
143
I'm not very familiar with matlab matrix manipulation, i'm sure i'm missing a clue somewhere !
thanks again for your help
  1 Kommentar
Walter Roberson
Walter Roberson am 16 Mär. 2011
It is not possible to have a numeric matrix with empty spaces.

Melden Sie sich an, um zu kommentieren.

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by