Aligning columns in a matrix by padding with zeros

I have a matrix A below. The values in each column are all roughly 100 apart, but some of them are missing some values. They represent loci of peaks in a graph. I want to make a matrix B that puts the zeros where they need to be: where a value in a column is "skipped"

1698	1696	1699
1826	1819	1816
1928	1921	1921
2063	2065	2059
2184	2186	2181
2399	2278	2270
2544	2373	2373
2655	2449	2555
2792	2552	2652
0	2657	2796
0	2794	0

e.g. column 1 should be

1698
1826
1928
2063
2184
0
2399
0
2544
2655
2792

[KSSV]

How you will determine that a column is skipped?

[AbioEngineer]

Hi KSSV, I'm new to machine learning and am not sure how to determine if a value in a column is skipped. I guess maybe I can find the column with the most values, and and compare each row value of a column with the distributed values of each row?

Answer 1

[Stephan]

Under the assumption that you have at least one not corrupted column (here column 2), this might work:

A = [1698 	1696	1699
1826	1819	1816
1928	1921	1921
2063	2065	2059
2184	2186	2181
2399	2278	2270
2544	2373	2373
2655	2449	2555
2792	2552	2652
0	2657	2796
0	2794	0]

non_corrupt_col = 2;
b = fix(A/100) - fix(A(:,non_corrupt_col)/100);

while sum(b==1,'all') ~=0
    b = fix(A/100) - fix(A(:,non_corrupt_col)/100);
    [r,c] = find(b==1,1);
    A(r+1:size(A,1),c) = A(r:size(A,1)-1,c);
    A(r,c) = 0;
end

disp('result:')
A

Results in:

result:

A =

        1698        1696        1699
        1826        1819        1816
        1928        1921        1921
        2063        2065        2059
        2184        2186        2181
           0        2278        2270
        2399        2373        2373
           0        2449           0
        2544        2552        2555
        2655        2657        2652
        2792        2794        2796