cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
The Discovery Summit 2025 Call for Content is open! Submit an abstract today to present at our premier analytics conference.
Choose Language Hide Translation Bar
Tntj
Level II

Formula for FPC elements in functional data explorer

I have a question regarding the FPC elements in the Functional Data Explorer. Is there any way that I can extract the formula for those elements? I can only see the formula for the shape functions when I extract the FPCA Model summaries. 

3 REPLIES 3

Re: Formula for FPC elements in functional data explorer

Hi,

It isn't completely straight forward but can be done in a few steps:

 

1) First, extract the formula by first saving the summaries:

Elodie_Delclaux_0-1677076930006.png

 

 

2) From the summary table, you can find all of the Shape Function formulas within the hidden columns of the table:

Elodie_Delclaux_1-1677076930023.png

 

3) Clicking on the + symbol will open the Formula editor for that column, showing the formula.  Double-clicking on it will display it in JSL, which can then be selected and copied:

Elodie_Delclaux_2-1677076930028.png

 

 

4) Next, create a new column in your original table, or in a new table with the independent variable in another column (in my case, Time):

Elodie_Delclaux_3-1677076930037.png

 

 

5) You can now explore the function in more detail, for example, by plotting it in Graph Builder if you wanted:

Elodie_Delclaux_4-1677076930041.png

 

 

Does this do what you were hoping for?

 

Tntj
Level II

Re: Formula for FPC elements in functional data explorer

 Thanks for your reply Elodie. But my question was mainly about the FPC components and the formula to calculate them if I know the shape functions. I have a case that I want to extract the shape functions from some historical data and then use that shape functions to calculate the FPC elements of a new batch. Is there any way around it?

ih
Super User (Alumni) ih
Super User (Alumni)

Re: Formula for FPC elements in functional data explorer

Unfortunately I do not know of an easy or straight forward way to do this, although I have many similar use cases.  

 

@Bill_Worley this fits in with our conversation.