How do I create an array containing the names of all the variables in a Workspace and then use this array to run a function/script on all the variables?

I have a workspace with about 200 tables. I frequently need to run similar operations (scripts) on all the tables.

1. First, is there a way to create an array that automatically pulls the names of the tables?
2. Second, no matter if the first one is possible or not; if I have a cell array containing all the names of the tables is there a way that can call these names and apply operations/scripts on them.

For example If I have the tables x_1, x_2, x_3 ... , x_200 in the workspace.

 var_names = {'x_1'; 'x_2'; 'x_3' ... 'x_200'}; %is there an easier way to do this.

 % if I want for example to multiply the first column in the three variable, I currently have to do it like this
 %
 x_1{:, 1}= x_1{:, 1} * 100;
 x_2{:, 1}= x_2{:, 1} * 100;
 x_3{:, 1}= x_3{:, 1} * 100;
 ...
 x_200{:, 1}= x_200{:, 1} * 100;
 %
 % is there anoter way to use a loop to call all the variables using the var_names array? It should in principle look like this.
 %
 for n = 1:200;
    "something(n)" = "something(n)" * 100;
 end

Thanks in advance.

Best

Antonius

[per isakson]

"table" &nbsp table is a special data type in Matlab, see table, Create table from workspace variables

Regarding your question see

• TUTORIAL: Why Variables Should Not Be Named Dynamically (eval) by Stephen Cobeldick
• FAQ: How Can I Create Variables A1 A2 ... A10 in a Loop by Jan Simon
• Alternatives to the eval Function

[Antonius Armanious]

The 200 variables in my workspace are in a table form; i.e., x_1, x_2, x_3 ... etc are all tables.

[Antonius Armanious]

Thanks for the links. I have seen most of them before. The problem is that my tables (variables) are not named sequentially as the example I gave, they have longer names to make it easy to identify which table refer to which experiment. These are examples of my tables (variables) names:

'fr_Au_pH7_I10_RAW_4624'

'fr_CYS_pH8_I50_RAW_4212'

'MS2_COOH_pH7_I10_RAW_2693'

Just to make it clear, I am not trying to name variables dynamically, I am trying to call my variables "dynamically'.

[per isakson]

I doubt that what you have are "table" in the Matlab meaning of the word. I guess, they are cell arrays.

[Antonius Armanious]

These are two screen shots which might help you better understand what I call table.

[per isakson]

• The answer to your first question is whos, List variables in workspace, with sizes and types
• Regarding the second question, we are reluctant to answer, because we don't want to promote bad practice.

[Stephen23]

@Antonius Armanious : don't do this:

https://www.mathworks.com/matlabcentral/answers/304528-tutorial-why-variables-should-not-be-named-dynamically-eval

[Antonius Armanious]

Thanks for helping me out here.

• First whos will list them and create a structure of all the information about the variable but will not create a cell array with the name, sorry if I am missing here.
• Second, I do not see why is this a bad practice. I understand that naming variables dynamically is a bad practice, but that is not the case. Can you suggest a data structure other than the one I have at the moment, which will allow me to both

1. recognize the experiments by their name (variable names); I can put all these tables in one cell array, but I won't be able to identify which experiment is which.
2. manipulate them collectively without having to call them one by one.

[per isakson]

Okay, you have tables. However, the syntax of &nbsp x_3{:, 1}= x_3{:, 1} * 100; &nbsp still confuses me.

[Antonius Armanious]

Forget about this, it was just an example. In my real script, that would look like

fr_CYS_pH7_I10_RAW_4133.Time = fr_CYS_pH7_I10_RAW_4133.Time/60;

[Walter Roberson]

Assign the output of whos to a variable. The result will be a structure array that has a name field. You can use structure expansion to tosd the content of the fields into one cell array.

Answer 1

[Walter Roberson]

http://www.mathworks.com/matlabcentral/answers/304528-tutorial-why-variables-should-not-be-named-dynamically-eval

[Antonius Armanious]

Thanks, this is the same answer I had for per up here.

Thanks for the links. I have seen most of them before. The problem is that my tables (variables) are not named sequentially as the example I gave, they have longer names to make it easy to identify which table refer to which experiment. These are examples of my tables (variables) names:

'fr_Au_pH7_I10_RAW_4624'
'fr_CYS_pH8_I50_RAW_4212'
'MS2_COOH_pH7_I10_RAW_2693'

Just to make it clear, I am not trying to name variables dynamically, I am trying to call my variables "dynamically'.

[Walter Roberson]

The line

      "something(n)" = "something(n)" * 100;

is an assignment statement. You are trying to name your variables dynamically.

"they have longer names to make it easy to identify which table refer to which experiment. "

Use a structure of tables and dynamic field names if you need that.

[per isakson]

Here you partly answer my questions. I missed it, because I didn't scroll down until now.

[Stephen23]

@Antonius Armanious: creating or accessing variable names dynamically is a bad idea. It does not matter if you are creating them or accessing them, they are both equally slow and buggy. You do not tell us how you got those variable into your workspace, but most likely through some importing, in which case there are much better alternatives (hint: indexing). Read these to know more:

https://www.mathworks.com/matlabcentral/answers/304528-tutorial-why-variables-should-not-be-named-dynamically-eval#answer_236122

https://www.mathworks.com/matlabcentral/answers/143-how-do-i-make-a-series-of-variables-a1-a2-a3-a10

Beginners often come here and tell us how much they need to define/access variable names dynamically, and how this will solve all of their problems. It won't. In fact this is what the MATLAB documentation says "A frequent use of the eval function is to create sets of variables such as A1, A2, ..., An, but this approach does not use the array processing power of MATLAB and is not recommended. The preferred method is to store related data in a single array."

https://www.mathworks.com/help/matlab/matlab_prog/string-evaluation.html

Experienced users and the official MATLAB documentation are telling you that accessing variable names dynamically is a bad idea. Do you think that with their combined millions of hours of experience and diverse backgrounds, they might have some good reason for all agreeing on this?

Storing meta-data in variable names is also a very bad practice, although very popular with beginners. Basically when you store meta-data in a variable name you make accessing the meta-data and the variable names difficult, as you are discovering right now. It is much much simpler to store your meta-data as data in its own right in their own variables. Then accessing your main data and your meta-data becomes trivial:

https://www.mathworks.com/matlabcentral/answers/225435-save-variable-as-string-from-user-input

You may be interested in structures, which are useful where data is complex:

http://www.mathworks.com/help/matlab/matlab_prog/generate-field-names-from-variables.html

Summary: do not put your meta-data in the variable names. Store the meta-data in its own variables (or as table header/rows). Then your code will be simpler, neater, faster, and less buggy. The real solution is to not import your data as lots of separate variables in the workspace: this is usually pretty trivial, e.g. load into a structure.

Question: how did those variables get into the workspace?

[Antonius Armanious]

@Walter Roberson and @per isakson

There must be something I missing here or I did not clearly explain what I am trying to do.

In this line what I am trying to do is to give an example of calling all the variables/tables in my workspace, which all a has unique names (and these names are essential to make them identifiable) and run an operation on them all.

   "something(n)" = "something(n)" * 100;

In the real script that looks something like this.

 fr_CYS_pH7_I10_RAW_4133.Time = fr_CYS_pH7_I10_RAW_4133.Time/60;

I will look into your suggestion of "structure of tables and dynamic field names" which I am not sure what it is and how it will look like.

[Antonius Armanious]

@Stephen Cobeldick

Thanks for the long reply. I have some reading to do before I can understand your suggestion with saving the metadata because I contemplated different approaches than putting it in the variable name but I found no clear alternative. I have to read through the links you provided to be able to judge if there is a good alternative or not.

Answer: These were experiment generated over years and every time I just imported the experiment in a new table/variable.

[Walter Roberson]

You would not use fr_CYS_pH7_I10_RAW_4133.whatever and MS2_COOH_pH7_I10_RAW_2693.whatever . You would use, for example, Ts.fr_CYS_pH7_I10_RAW_4133.whatever and Ts.MS2_COOH_pH7_I10_RAW_2693.whatever . You would then be able to do

fn = fieldnames(Ts);
for K = 1 : length(fn)
   this_table = fn{K};
   Ts.(this_table).Time = Ts.(this_table).Time / 60;
end

[Stephen23]

@Antonius Armanious: I know how it is, when experimental data is stored with some meta-data in the filename:

 test_80kmh_25degC_900amps.csv

That may be fine for storing the data in files. This is intuitive and obvious to humans when creating those files, but when importing that data into MATLAB (or most other programming languages) it would be a poor practice to try and keep those names as variable names (for the reasons give in our comments, and the links I gave). Instead, that meta-data should be parsed and imported as data in its own right (which makes it easier to work with too!), and all of the measured data imported into one simple variable using indexing (or table rows, fieldnames, etc).

[Antonius Armanious]

Thanks everybody for helping me out here. That is very much appreciated.

Answer 2

[per isakson]

I guess you got the message. (I might delete this answer.) Try this, which is a solution if you painted yourself into a corner.

T = table(categorical({'M';'F';'M'}),[45;32;34],...
    {'NY';'CA';'MA'},logical([1;0;0]),...
    'VariableNames',{'Gender' 'Age' 'State' 'Vote'});

fr_Au_pH7_I10_RAW_9999      = T;
fr_CYS_pH8_I50_RAW_9999     = T;
MS2_COOH_pH7_I10_RAW_9999   = T;
sas = whos( '-regexp', '.+?RAW_\d{4}' );
sas = reshape( sas, 1,[] );

MyTables = struct('Promise', 'I will never do this again! ');

for s = sas
    MyTables.(s.name) = eval( s.name );
end
name_list = fieldnames( MyTables );
name_list(1) = [];
name_list = reshape( name_list, 1,[] );

for name = name_list
    MyTables.(name{:}) = foo( MyTables.(name{:}), 17 );
end

where

function    T = foo( T, N )
    T.Age = T.Age + N;
end

[Antonius Armanious]

Thanks, @per isakson for helping me out here. I will try everything else before I end up using this approach.

[per isakson]

You know how to decipher the names. I can guess what "pH8" stands for, but that's about all.

The screen clips indicate that you have a large amount of experimental data. I would have stored that in an HDF5-file. However, I recognize ... . But in the long run that might pay off.

A somewhat better structure than MyTables might be something like

S.DateTime
S.attribute_1
S.pH
S.attribute_2
S.Table

Then you create a 1D structure array with length equal to the number of tables. For selecting

is8 = [S.pH]==8; S8 = S(is8);

creates a subset with pH==8.

[Stephen23]

@Antonius Armanious: what per isakson showed is how much simpler it is to work with data when it is stored correctly: using some very simple syntax per isakson compared data, and selected a subset of that data based on the comparison. Using lots of variables with complicated names would make this task very complex, slow, and buggy.

That is why we have been telling you to use better data storage (i.e. variables): because it makes your code better!

[Antonius Armanious]

@per isakson and @Stephen Cobeldick. Thanks again. I just need to get myself familiarized with using structures to organize my data. I think I have a steep learning curve but I will follow your advice.