Script for standard deviation based on +/- 1.5 sigma

[#1]

Hi,

I like to calculate standard deviation based on looking at data with ranges of values (+/- 1.5 sigma). Is there anyone you can achiev that in JMP?

I appreciate your help. Thanks

[#2]

is this what you want>

Names Default To Here( 1 );
dt = Open( "$SAMPLE_DATA/semiconductor capability.jmp" );

sigma = Col Std Dev( :npn1 );
special std dev = Col Std Dev( If( (Col Mean( :npn1 ) - 1.5*sigma) <= :npn1 | :npn1 >= (Col Mean( :npn1 ) + 1.5*sigma), ., :npn1 ) );
Show( sigma, special std dev );

[#3]

Thanks Jim for quick feedback and solution. This what I want. I lalso

---

@txnelson wrote:

is this what you want>

Names Default To Here( 1 );
dt = Open( "$SAMPLE_DATA/semiconductor capability.jmp" );

sigma = Col Std Dev( :npn1 );
special std dev = Col Std Dev( If( (Col Mean( :npn1 ) - 1.5*sigma) <= :npn1 | :npn1 >= (Col Mean( :npn1 ) + 1.5*sigma), ., :npn1 ) );
Show( sigma, special std dev );

---

ike to do this for many columns in a data table and then create a new colum with Pass or Fail values. I supose the spec limits for each column can be read from a data table and then new column with "Pass" or Fail" based on control limits.

If anyone has done this type of work, I will be intersted to learn from that. Thanks

[#4]

Hi Jim,

I followed your script and added list of columns to calculate this special sigma. When I run the script, I get

sigma = .;

special std dev = .;

sigma = .;

special std dev = .;

sigma = .;

Here is the script:

Names Default To Here( 1 );

dt = Open( "$SAMPLE_DATA/semiconductor capability.jmp" );

Current data Table();

lstNames = dt << Get Column Names(String,Numeric);

for (i =1, i <= N Items (lstNames), i++,

if (contains(lstNames[i],"NP") == 1,

sigma = Col Std Dev( lstNames[i]);

special std dev = Col Std Dev( If( (Col Mean( lstNames[i] ) - 1.5*sigma) <= lstNames[i] | lstNames[i] >= (Col Mean( lstNames[i] ) + 1.5*sigma),.,lstNames[i] ) );

Show( sigma, special std dev );

)

);

I appreciate your help. Thanks

---

@AT wrote:

Thanks Jim for quick feedback and solution. This what I want. I lalso

---

@txnelson wrote:

is this what you want>

Names Default To Here( 1 );
dt = Open( "$SAMPLE_DATA/semiconductor capability.jmp" );

sigma = Col Std Dev( :npn1 );
special std dev = Col Std Dev( If( (Col Mean( :npn1 ) - 1.5*sigma) <= :npn1 | :npn1 >= (Col Mean( :npn1 ) + 1.5*sigma), ., :npn1 ) );
Show( sigma, special std dev );

---

ike to do this for many columns in a data table and then create a new colum with Pass or Fail values. I supose the spec limits for each column can be read from a data table and then new column with "Pass" or Fail" based on control limits.

If anyone has done this type of work, I will be intersted to learn from that. Thanks

---

[#5]

The argument to Col Std Dev needs to be a column reference not the string name.  You can use the Column function to generate a column reference from the string name:

sigma = Col Std Dev( Column( lstNames[i]) );

[#6]

Thanks Dave.

[#7]

The issue you are running into is that you were placing into the formula not a column name, like NPN1, but rather a literal string of "NPN".  The Substitute function allows one to take the "NPN1" and substitute in :NPN1.

Names Default To Here( 1 );
dt = Open( "$SAMPLE_DATA/semiconductor capability.jmp" );
Current Data Table();

lstNames = dt << Get Column Names( string, Numeric );

For( i = 1, i <= N Items( lstNames ), i++,
	If( Contains( lstNames[i], "NP" ) == 1,
		sigma = Col Std Dev( Column( dt, lstNames[i] ) );
		Eval(
			Substitute(
					Expr(
						special std dev = Col Std Dev(
							If( (Col Mean( __col__ ) - 1.5 * sigma) <= __col__ | __col__ >= (Col Mean( __col__ ) + 1.5 * sigma),
								.,
								__col__
							)
						)
					),
				Expr( __col__ ), Parse( ":" || lstNames[i] )
			)
		);
		Show( sigma, special std dev );
	)
);

And here is another way to specify that the literal string is to be treated as a column name

Names Default To Here( 1 );
dt = Open( "$SAMPLE_DATA/semiconductor capability.jmp" );
Current Data Table();

lstNames = dt << Get Column Names( string, Numeric );

For( i = 1, i <= N Items( lstNames ), i++,
	If( Contains( lstNames[i], "NP" ) == 1,
		sigma = Col Std Dev( Column( dt, lstNames[i] ) );
		
		special std dev = Col Std Dev(
			If(
				(Col Mean( As Column( lstNames[i] ) ) - 1.5 * sigma) <= As Column( lstNames[i] ) | As Column( lstNames[i] ) >= (
				Col Mean( As Column( lstNames[i] ) ) + 1.5 * sigma),
				.,
				As Column( lstNames[i] )
			)
		);
	);
	Show( sigma, special std dev );
);

[#8 · Accepted solution]

Thanks Jim. It is very helpful.