Find Character Patterns in Data Set

robot

Hi,

I have a data set that lists widgets in one column and a long string of defect codes in another column.  Is there a platform in JMP well suited to find character patterns?  For example, assuming defect codes A-Z, which may appear in any quantity per row, do the codes {A,E,F} often appear together?  Or maybe sometimes the codes {H,L,P,X} often appear together?  I am using JMP10.  Thanks!

Example Table

nt = New Table( "Defect Codes",
  Add Rows( 3 ),
  New Column( "Widget ID",
  Numeric,
  Continuous,
  Format( "Best", 12 ),
  Set Values( [1, 2, 3] )
  ),
  New Column( "Defect Codes",
  Character,
  Nominal,
  Set Values( {"K,V,Q,D", "A,P", "N,D,H"} )
  )
);

ms

I agree with michael that this is an interesting problem. One idea is to expand the error codes over multiple columns (lets say columns A to Z) and perform a missing data pattern (Tables Menu). The pattern may directly be useful for illustrating associations between defect codes, but an additional step could be to apply Multivariate() on the missing pattern. Or even som flavour of the categorial platform.

A jsl example:

//list of all defect codes

def_code_list = {"A", "B", "C", "D", "E", "F", "G", "H", "I", "J", "K", "L", "M", "N", "O",

"P", "Q", "R", "S", "T", "U", "V", "W", "X", "Y", "Z"};

n = N Items( def_code_list );

//–––––––––––––––––––––––

// Example table

nt = New Table( "Defect Codes",

  Add Rows( 200 ),

  New Column( "Widget ID", Numeric, Continuous ),

  New Column( "Defect Codes", Character, Nominal, )

); 

Column( 1 ) << Set Values( Index( 1, N Row( nt ) ) );

//Add random effect codes

For Each Row(

  :Defect Codes = Substitute(

  Char( def_code_list[Random Index( n, Round( Random Exp() + 2, 0 ) )] ),

  "{", "",

  "}", "",

  "\!"", ""

  )

);

//contaminate with some common combinations

:Defect codes[Random Index( 200, 10 )] = "A, B, C";

:Defect codes[Random Index( 200, 20 )] = "M, N, O";

// Expand to multiple columns (and make column list)

collist = {};

For( i = 1, i <= n, i++,

  col = nt << New Column( def_code_list[i], character, nominal );

  Insert Into( collist, col );

);

For Each Row(

  ID_codelist = Words( :Defect Codes[], ", " );

  For( i = 1, i <= N Items( ID_codelist ), i++,

  Column( nt, Loc( def_code_list, ID_codelist[i] )[1] + 2 )[] = ID_codelist[i]

  );

);

//––––––––––––––––––––––––

// Analyze data

// Missing data pattern

mdp = nt << missing data pattern( columns( Eval( collist ) ) );

// Multivariate on missing pattern

mdp << multivariate( columns( Eval( collist ) ), freq( :Count ) );