470274_en-US

liliconcom

Script to loop thru and group columns by name, dynamically

So I'm trying to figure this out and its just not quite working.

 

I've got a table with a key-value pair that I've SPLIT from Tall to Wide. I end up with about 10k columns that all have a regular naming structure, like 
"15.02.0 param1", "15.02 param2", "15.02 param3"...."15.03 param1", "15.03 param2", "15.03 param3",...

 

How can I look through all these columns and group them dynamically by the starting term? ie. all 15.02% columns in a group, all 15.03% cols, all 15.04%, ... all 17.06%....and name the group based on the groupby number?

 

My script just groups everything into a single group. ug.

 

//**split Aggregate Data from TALL to WIDE, grouping WIDE columns by StepID
dt = current data table();
Summarize( dt, RecipeStepIndex = by( :RecipeStepId) );//get list of unique RecipeStepID's

//split the table TALL to WIDE
dt << Split(
       Split By( :RecipeStepId, :ParameterName, :AggregateFunction ),
       Split( :Value ),
       Group( :Recipe, :ProcessTime, :Tool, :Chamber, :Lot, :WaferScribe ),
       Output Table( "AggWide" ),
       Remaining Columns( Drop All ),
       Sort by Column Property,
       dt << sort( by( :Recipe, :EquipmentId, :ChamberId, :LotId, :SlotNo ))
);

dtAggWide = current data table();
ColNamesList = dtAggWide << get column names(string, continuous);  //get the column names of the WIDE table
//Create an empty list for the matching columns
RecipeStepList = {};
// Find matching column names
For( i = 1, i <= N items( RecipeStepIndex ), i++,
 For( j = 1, j <= N Items( ColNamesList ), j++,
  If( 
   Contains( ColNamesList[j], RecipeStepIndex[i] ),
    Insert Into( RecipeStepList, ColNamesList[j] ),      
  );
 );
);
// If groups of columns are found, create Col groupings
If( N Items( RecipeStepList ) > 0,
 dtAggWide << select columns( RecipeStepList );
 dtAggWide << group columns();
);

ScriptingRe: Script to loop thru and group columns by name, dynamically

There are many ways to do this.

 

Here is one example which first builds the groups into associative array with the key as name of the group and value is list of columns in the group, then loops over that associative array to group columns in the data table. To get group names we use Word() function on the column names. This requires JMP16 due to using For Each

 

Names Default To Here(1);
// this solution requires JMP16

dtAggWide = New Table("Untitled",
 Add Rows(0),
 New Column("15.02.0 param1", Numeric, "Continuous", Format("Best", 12), Set Values([])),
 New Column("15.02.0 param2", Numeric, "Continuous", Format("Best", 12), Set Values([])),
 New Column("15.02.0 param3", Numeric, "Continuous", Format("Best", 12), Set Values([])),
 New Column("15.03 param1", Numeric, "Continuous", Format("Best", 12), Set Values([])),
 New Column("15.03 param2", Numeric, "Continuous", Format("Best", 12), Set Values([]))
);

col_list = dtAggWide << Get Column Names("String");

// first get the column group names and columns<BR />// in this solution case we use associative array
aa_groups = Associative Array();
For Each({col_name}, col_list, 
 col_group = Word(1, col_name, " ");
 If(!Contains(aa_groups, col_group),
  aa_groups[col_group] = {};
 );
 Insert Into(aa_groups[col_group], col_name)
);
Show(aa_groups);

For Each({{key, value}}, aa_groups,
 dtAggWide << Group Columns(key, value); 
);

 

Re: Script to loop thru and group columns by name, dynamically

thankyou, will try this shortly.

But what about building the grouping array list dynamically? Basically, each unique value of :RecipeStepId should be a groupby name. I tried doing that at the beginning, before splitting the table, saving it to the "RecipeStepIndex"

Summarize( dt, RecipeStepIndex = by( :RecipeStepId) );//get list of unique RecipeStepID's

 

 

Re: Script to loop thru and group columns by name, dynamically

You could "pre-fill" the Associative array for example with the column names based on RecipeStepIndex. Most likely RecipeStepIndex will be found from the start of split column names so this should still work.

 

 

Names Default To Here(1);
// this solution requires JMP16

dtAggWide = New Table("Untitled",
 Add Rows(0),
 New Column("15.02.0 param1", Numeric, "Continuous", Format("Best", 12), Set Values([])),
 New Column("15.02.0 param2", Numeric, "Continuous", Format("Best", 12), Set Values([])),
 New Column("15.02.0 param3", Numeric, "Continuous", Format("Best", 12), Set Values([])),
 New Column("15.03 param1", Numeric, "Continuous", Format("Best", 12), Set Values([])),
 New Column("15.03 param2", Numeric, "Continuous", Format("Best", 12), Set Values([]))
);

col_list = dtAggWide << Get Column Names("String");

// first get the column group names and columns// in this solution case we use associative array
aa_groups = Associative Array();

//Summarize(dt, RecipeStepIndex = by(:RecipeStepId));//get list of unique RecipeStepID's
RecipeStepIndex = {"15.02.0", "15.03"};
aa_groups = Associative Array(RecipeStepIndex, Repeat({{}}, N Items(RecipeStepIndex)));

For Each({col_name}, col_list, 
 col_group = Word(1, col_name, " ");
 Insert Into(aa_groups[col_group], col_name)
);

For Each({{key, value}}, aa_groups,
 dtAggWide << Group Columns(key, value); 
);