cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
Check out the JMP® Marketplace featured Capability Explorer add-in
Choose Language Hide Translation Bar
Cactus123456
Level I

Using a formula column, how do I find a value in a subset of the data based on other columns

If I have two different time series in a table, and there is a findable event in both series, but they occur at different times. How can I specify a formula column to subtract off the time such that the finable events start at time zero?

example.PNG

1 ACCEPTED SOLUTION

Accepted Solutions
hogi
Level XII

Re: Using a formula column, how do I find a value in a subset of the data based on other columns

If the signal is very clean, you can just use lag to find the jump - otherwise you have to invest more effort to remove noise etc.

Then you just have to subtract the offset of the jump from your original time.

Col ... aggregations like Col Maximum can be used to spread one value to all rows - which then can be subtracted.

A second argument of the Col ... aggregation can be used as GroupBy argument (1 maximum per group) to subtract individual offsets for the 2 traces.

 

hogi_1-1698356460779.png

 

t1 = random integer(1000);
t2 = random integer(1000);

dt = new table("test",
add rows(2000),
New Column( "time",  Set Each value(Mod(row()-1,1000)) ),
New Column( "sample",  Set Each value(floor((row()-1)/1000)+1) ),

New Column( "trace",  Set Each value(if(:time<match(:sample,1,t1,2,t2),1,0 )))
);

New Column( "marker",  Set Each value(if(lag(trace)==1 & :trace==0,time,.)));

New Column( "time_adjusted",
	set each value( :time - Col Maximum( :marker, :sample ) ),
	Set Selected
);

Graph Builder(
	Variables(
		X( :time ),
		X( :time_adjusted ),
		Y( :trace ),
		Group Y( :sample )
	),
	Elements( Position( 1, 1 ), Line( X, Y, Legend( 14 ) ) ),
	Elements( Position( 2, 1 ), Line( X, Y, Legend( 15 ) ) )
)

 

View solution in original post

2 REPLIES 2
hogi
Level XII

Re: Using a formula column, how do I find a value in a subset of the data based on other columns

If the signal is very clean, you can just use lag to find the jump - otherwise you have to invest more effort to remove noise etc.

Then you just have to subtract the offset of the jump from your original time.

Col ... aggregations like Col Maximum can be used to spread one value to all rows - which then can be subtracted.

A second argument of the Col ... aggregation can be used as GroupBy argument (1 maximum per group) to subtract individual offsets for the 2 traces.

 

hogi_1-1698356460779.png

 

t1 = random integer(1000);
t2 = random integer(1000);

dt = new table("test",
add rows(2000),
New Column( "time",  Set Each value(Mod(row()-1,1000)) ),
New Column( "sample",  Set Each value(floor((row()-1)/1000)+1) ),

New Column( "trace",  Set Each value(if(:time<match(:sample,1,t1,2,t2),1,0 )))
);

New Column( "marker",  Set Each value(if(lag(trace)==1 & :trace==0,time,.)));

New Column( "time_adjusted",
	set each value( :time - Col Maximum( :marker, :sample ) ),
	Set Selected
);

Graph Builder(
	Variables(
		X( :time ),
		X( :time_adjusted ),
		Y( :trace ),
		Group Y( :sample )
	),
	Elements( Position( 1, 1 ), Line( X, Y, Legend( 14 ) ) ),
	Elements( Position( 2, 1 ), Line( X, Y, Legend( 15 ) ) )
)

 

Cactus123456
Level I

Re: Using a formula column, how do I find a value in a subset of the data based on other columns

This solved my problem exactly!

 

I think col ... aggregations was something I was sorely missing in my mental toolbox.

 

Thank you!