Shell Programming and Scripting

Hey guyz,

I have a table which shows the presence or absence of my variables (A,B,C,...) in my observations (1,2,3,...)


I want to calculate the co-presence of my variables. to have a table shows the pairwise presence of the variables (have value 1). something like this:


You see in this table there are 3 observations which variable A is present and for A&B there is one but for example for B&C there is no co-presence and so far.

I would be so grateful if you can help me with this.

Well, the SQL flavored solution is to decompose the table into a simpler table, columns into rows, with a union in a derived table, for instance:

Code:

Var Obs
A 1
A 2
A 3
B 2
C 1

Now you can join that table to itself, where a.obs=b.obs and (to avoid self and duplicate corelations, if you care) a.var < b.var

In SQL, when someone puts the cells of like entities that should be many rows into columns, a rotated array, you need unions to view the few rows many columns as one column many rows. Conversely, if you need to turn cells from many rows into columns of one row, you need to use joins. It is a common RDBMS design error to put arrays horizontally in rows when they belong vertically in one column many rows.

Now if we start talking lines and columns, 'union' is concatenate (cat, >> or just multiple passes in a subshell), 'join' is, surprise, 'sort' and 'join'; 'where' is in the join or grep. Deciding 'A' < 'B' is more a perl chr()/ord() or C question, but you were accepting self and duplicate matches above.