Convert workspace of cells to double

Hi I have a workspace which is full of cell arrays I want a way which can change the entire workspace to double array.

rite now my workspace is small so i am able to use cell2mat(variable name). But there is a good chance it will increase in the future. Can anyone help me with it.

[Guillaume]

While it's certainly possible to do what you want, in my opinion a better answer would be to remove the need to do it in the first place.

Why have you got a workspace full of cell arrays that you all want to convert? If all these cell arrays had been created inside a container then the conversion would be trivial? So can you explain where these cell arrays come from?

[sc1991]

Well, i had to do some processing based on an excel sheet the problem was the header had all the variables in Caps and with few special symbols but i needed all in small cases without any special symbols. So,I made a excel sheet with the correct variable name in one column and all Caps variable in the second column. Now my script picked the correct name from the column and went back to main excel sheet which has all the names in Caps and just took the data and assigned it to my variable name (without caps one.). Well now all this is done but during this whole processing thing it just converted all these variables to cell arrays.

[Stephen23]

Use a structure and the whole problem will go away.

Or put the data into one numeric array.

Solving problems through better data and code design is much preferred to writing slow, complex, buggy code that tries to access all variables in the workspace.

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

"rite now my workspace is small so i am able to use cell2mat(variable name). But there is a good chance it will increase in the future."

You have described a perfect example of how different paradigms apply as program complexity or data scales up: what works well for two or three variables becomes intractable for one million variables. There is nothing wrong with that, it just means that it is time to change what your code does!

"Can anyone help me with it."

Yes, you can. By not spamming lots of variables into the workspace.

[sc1991]

I mean i did think of using structures but i have never worked on them so i was a bit hesitant. and apart from this i wont be using this code. This code will go to someone who just will have to update the Excel sheet with correct name and run my script. So in future they might add up around i m guessing 500 variables. which my script can take care and help them get their result its just that final part of conversion from cell 2 double. But let me see about implementing structures. Thanks for the suggestion

[sc1991]

So this is my code.

 clc
clear
[~, ~, raw] = xlsread('varnames.xlsx','Sheet1');%varnames is the name of the Excel doument which has real name and alias name column
AllAliasPara=raw(2:end,1);
Realparam=raw(2:end,2);
RealParaIdx=find(cell2mat(cellfun(@(x)any(~isnan(x)),Realparam,'UniformOutput',false)));%Sometimes you have an Alias name but you don have a real var name so the real name row gets a NaN
AliasParaAvailable=AllAliasPara(RealParaIdx);%Available alias name 
Realparanameavailable=Realparam(RealParaIdx);%Available Real name 
xlswrite('Realparanameavailable.xlsx',Realparanameavailable);
[mydata, myheader] = xlsread('Realparanameavailable.xlsx');
[~,~,raw1]=xlsread('dataset.xlsx');
ii=0;
for iii=1:length(AliasParaAvailable)
     colnum=strcmp(AliasParaAvailable(iii),raw1(1,:));
     dataset=raw1(3:end,colnum);
     mydata=dataset;
     ii=ii+1;
         for ii=ii
            commandExec = [myheader{ii}, ' = ', 'mydata(:,', num2str(1) , ');'];
            evalin('base', commandExec );
         end     
end

clearvars ii iii colnum dataset raw raw1 commandExec mydata myheader AllAliasPara Realparam RealParaIdx AliasParaAvailable Realparanameavailable

and the following are sample Excel documents I have created

Real Name  Alias Name
 Temp            _TEMP@
 Engine_Speed    _ENGINE_SPEED

This is the Excel sheet Varnames.

_TEMP@  _ENGINE_SPEED
deg C  rpm
35  700
36  800
37  850
40  900
This is the Excel sheet Data set.

My task is to take the data from data set Excel sheet and use the varnames Excel sheet to get the correct names of the Variable and store the value in it.

This is the code I have been working on. I am sure there is a much better way but rite now as I lack my knowledge in structures I can only think of using eval.

Another concern is by using the above script i get all the variables as Cell and for me to work on them any further i need to convert them to double. I am not sure how to do that.

I have read your comments but I am still not clear how to do it. So please suggest what changes i need to make.

[Stephen23]

"...rite now as I lack my knowledge in structures I can only think of using eval"

So you want to use eval... and you immediately get stuck on the trivial problem of accessing your data that you can't solve so you have to ask strangers to help you. Do you see the connection yet?

Accessing data is a trivial functionality, it should not cause problems like this. Importing data should be reliable, because the rest of your code depends on it. Why waste time messing around with eval or evalin when presumably actually processing your data is what you want to be doing?

"I am sure there is a much better way"

There are. Perhaps this would be a pertinent time to consider writing better (simpler, faster, neater, more efficient, much more reliable) code?

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

Answer 1

[Guillaume]

We keep on saying that eval and co. are bad practice for good reasons. Yes, it may make it easy to create all these variables that you can easily see in your workspace with recognisable names but ultimately, it just make it harder to use these variables programmatically. As you've found out when you want to convert all of them. So really, don't do it! Ultimately, you'll find that the alternative are easier to use.

Changing your code to create a structure is easy. Replace the code in the (useless) ii loop by:

somestructname.(myheader{ii}) = mydata(:, 1);

However, assuming that you're using a fairly recent version of matlab all your problems (dynamic names, cell array, renaming) would mostly go away if you used readtable to read your excel files:

aliases = readtable('varnames.xlsx', 'Sheet', 'sheet1'); %import alias spreadsheet
aliases(ismissing(aliases.RealName), :) = [];  %remove empty rows (if any)
aliases.importname = matlab.lang.makeValidName(aliases.AliasName);  %actual variable name when imported

opts = detectImportOptions('dataset.xlsx');
[match, index] = ismember(aliases.importname, opts.VariableNames);  %find real name of columns
opts.VariableNames(match) = aliases.RealName(index(match)); %replace column name
if ~all(match)
   warning('Some columns do not have aliases: %s', strjoin(opts.VariableNames(~match), ', '));
end
opts.SelectedVariableNames = opts.VariableNames(match); 
opts.VariableUnitsRange = 'A2';  %to import the unit row
dataset = readtable('dataset.xlsx', opts');  %automatically import with correct column name and type

[sc1991]

Thanks a lot. Actually I did fix my code last nite. But i have started reading about structures and I think your code has really shown me why I should really invest in structures and thanks for informing about readtables I honestly didn knew we could do that I will even read that. Thank you sooo much.

[sc1991]

clc
clear
[~, ~, raw] = xlsread('varnames.xlsx','Sheet1');%varnames is the name of the Excel doument which has real name and alias name column
AllAliasPara=raw(2:end,1);
Realparam=raw(2:end,2);
RealParaIdx=find(cell2mat(cellfun(@(x)any(~isnan(x)),Realparam,'UniformOutput',false)));%Sometimes you have an Alias name but you don have a real var name so the real name row gets a NaN
AliasParaAvailable=AllAliasPara(RealParaIdx);%Available alias name 
Realparanameavailable=Realparam(RealParaIdx);%Available Real name 
xlswrite('Realparanameavailable.xlsx',Realparanameavailable);
[mydata, myheader] = xlsread('Realparanameavailable.xlsx');
[~,~,raw1]=xlsread('dataset.xlsx');
ii=0;
for iii=1:length(AliasParaAvailable)
     colnum=strcmp(AliasParaAvailable(iii),raw1(1,:));
     dataset=raw1(3:end,colnum);
     mydata=cell2mat(dataset);
     ii=ii+1;
         for ii=ii
            commandExec = [myheader{ii}, ' = ', 'mydata(:,', num2str(1) , ');'];
            evalin('base', commandExec );
         end     
end
clearvars ii iii colnum dataset raw raw1 commandExec mydata myheader AllAliasPara Realpara

This code is just for refrence for anyone who gets struck like me. It was just adding a cell2mat before dataset in this script. However, I would not encourage to use eval as it can end up consuming a lot of your time. So, please invest in structures but just in case anyone is struck and you end up on this page you can use my code for temporary. But, I would strongly suggest to go with the method Guillaume has suggested. Thank you

[Stephen23]

You should also avoid evalin, and just pass data properly as arguments.

The brutal clc,clear at the start is totally pointless cargo-cult programming.

[sc1991]

Stephen the clc and clear is because i just want nothing to present in the workspace and I have a reason for it as my script is just one small part of a big program. But due to data senstivity I was unable to share the whole thing

[Stephen23]

"the clc and clear is because i just want nothing to present in the workspace..."

Did you know that functions have an empty workspace when they are called (apart from input variables) ?

"...and I have a reason for it as my script is just one small part of a big program."

If your code really is "a big program" then using you should be writing functions anyway, not scripts. Then you can trivially avoid slow introspective programming methods like clear as well as enjoying other benefits such as faster code evaluation, independent workspaces, ease of debugging, and much improved debugging tools.

https://www.mathworks.com/matlabcentral/answers/228557-experts-of-matlab-how-did-you-learn-any-advice-for-beginner-intermediate-users

[sc1991]

yea good advice honestly this script is only to test a solution and later everything will be made into function but nice link good info. Thanks

[Guillaume]

While we're on the topic of the code you wrote

RealParaIdx=find(cell2mat(cellfun(@(x)any(~isnan(x)),Realparam,'UniformOutput',false)));

This seems aimed at detecting empty rows. Why is there empty rows in the spreadsheet? Ignoring that, the cell2mat would be unnecessary if you hadn't ask cellfun to output a cell array, so:

RealParaIdx = find(cellfun(@(x)any(~isnan(x)),Realparam));

The find is also unnecessary. You can directly use the logical array returned by cellfun as a filter:

RealParaIdx = cellfun(@(x)any(~isnan(x)),Realparam)
AliasParaAvailable=AllAliasPara(RealParaIdx);  
Realparanameavailable=Realparam(RealParaIdx);

Then, I don't understand why you write an excel file ( Realparanameavailable.xlsx) to read it back on the next line.

Then there is of course

for ii = ii

which is completely pointless.

Finally, If you're going to build string by concatenation and conversion, I'd recommend using sprintf:

commandExec = sprintf('%s = mydata(:, %d);', myheader{ii}, 1);

But really, I'd recommend using tables with a code similar to what I've written.

[sc1991]

I am sorry it was a very poor code. I did make some change to it. I will post the code for now but I am going to redo the whole code using structure and the way you have suggested.

clc
clear
[~, ~, raw] = xlsread('varnames.xlsx','Sheet1');
AllAliasPara=raw(2:end,1);
Realparam=raw(2:end,2);
RealParaIdx = cellfun(@(x)any(~isnan(x)),Realparam);
AliasParaAvailable=AllAliasPara(RealParaIdx);
Realparanameavailable=Realparam(RealParaIdx);
[~,~,raw1]=xlsread('datset.xlsx');
for iii=1:length(AliasParaAvailable)
     colnum=strcmp(AliasParaAvailable(iii),raw1(1,:));
     dataset1=raw1(3:end,colnum);
     mydata=cell2mat(dataset1);
     eval(sprintf('%s=mydata(:,1);',Realparanameavailable{iii}));         
end
clearvars ii iii colnum dataset raw raw1 mydata myheader AllAliasPara Realparam RealParaIdx AliasParaAvailable Realparanameavailable