Splitting Table Based on One Column Value

Hello,

I have a 1790x34 table of values called dP, and I would like to create two other tables based on a variable called 'A_VAL'. If the value under the 'A_VAL' variable is less than 0.1, then it is stored in the 'lo_val' table, but it is stored in the 'hi_val' table if it is greater-than or equal-to 0.1. I have tried the following code:

idx = dP(:,'A_VAL') < 0.1;

lo_val = dP(idx,:)
hi_val = dP(~idx, :)

But, unfortunately, it gives me the following error:

Error using  () 
A table row subscript must be a numeric array containing real positive integers, a logical array, a
character vector, a string array, a cell array of character vectors, or a pattern scalar.

Error in Hedwig_HS_input (line 14)
lo_val = dP(idx,:)

Any assistance would be greatly appreciated, thank you!

Answer 1

[Stephen23]

Use the correct type of indexing. Curly braces returns the table content, not another table:

https://www.mathworks.com/help/matlab/matlab_prog/access-data-in-a-table.html

idx = dP{:,'A_VAL'} < 0.1;

For example:

T = array2table(rand(5,3))

T = 5×3 table
      Var1        Var2        Var3  
    ________    ________    ________

     0.50764     0.41567     0.52306
     0.84056     0.59901     0.46768
      0.7887     0.82771    0.079556
    0.031295    0.046708      0.6444
     0.41241     0.46735      0.3585

Look at the class of this: is it logical? Can it be used for indexing?

idx = T(:,'Var1') < 0.5 % what you did

idx = 5×1 table
    Var1 
    _____

    false
    false
    false
    true 
    true 

idx = T{:,'Var1'} < 0.5 % What you should have done, returns a logical array

idx = 5×1 logical array
   0
   0
   0
   1
   1

You could have diagnosed this yourself: always start debugging by looking at your data. The error message tells you that the indexing array must be numeric, logical, etc... so this is where you take the initiative to learn what the class is... simply printing it out is a good start. In contrast, not looking at your data won't debug much at all.

Indexing is a MATLAB superpower: understanding the differences between curly-braces and parentheses is critical to using indexing: curly-braces gives the array content, parentheses the array itself.

The syntax you used (arithmetic/boolean operation on a table directly) always returns a table:

https://www.mathworks.com/help/matlab/matlab_prog/rules-for-table-and-timetable-mathematics.html

[Patrick]

This has fixed my entire problem. Thank you so much, Stephen! MVP indeed! :)

In addition, thank you for these helpful tips on diagnosing these problems. I've only recently started going through the basics, but I admit that I am learning a lot as I go through.

[Image Analyst]

@Patrick tables are sometimes like cell arrays. See the FAQ for how to figure out when to use curly braces, square brackets, or parentheses.

https://matlab.fandom.com/wiki/FAQ#What_is_a_cell_array?

[Patrick]

Thank you very much! This also helps me greatly.

[Peter Perkins]

Just to add some additional clarity:

In older versions of MATLAB, the following comparison would be an error. In recent versions (R2023a), it works:

T = array2table(rand(5,3));
Tidx = T(:,'Var1') < 0.5

Tidx = 5×1 table
    Var1 
    _____

    true 
    true 
    true 
    false
    true 

But as Stephen23 says, the result is a table, because the presumption is that you have all your data in a table so you must want to stay in a table. Stephen23 describes how to extract a numeric, on which < returns a logical, but it's also possble to do the above and then

Tidx.Var1

ans = 5×1 logical array
   1
   1
   1
   0
   1

Either is fine, which is better depends what you are doing.