Create a new variable when conditions are met

Hello All,

A question from a novice here that has been working on what no doubt many of you will consider a very simple problem - but it still eludes me!

In a nutshell, I am trying to achieve the following:

• write a function that will create a new variable when certain conditions are met when looking at an existing variable

Example

Temp is a matrix (177,2) with varying data. When the function [datagrab] is called, if a number appears in the second column that matches the conditions eg '1' (this is not a simple logical test as other integer numbers could be 2,3,4,5 etc.), then that whole row should be copied to a new variable 'G'. If the datagrab function is called again say using datagrab('2','M') - then all rows of data from temp that have the value 2 in the second column would be written to a new variable called 'M', and so on. I hope that makes sense!

I have been trying to just get it to work with '1' and 'G' for now, and I have this:

   function [outindex] = datagrab(1,G)
  n = 1;
  outarray = [];

for G = length(temp)
    if (temp(:,2) == '1'); %eg whatever value set as '1' for now
        outarray(n)=G;
        n = n+1;
    else end
end

end

but its just not working for me. So, if anyone could offer a solution for the whole problem, or just a fix to my code above then I will be most grateful.

Heres hoping!

10B.

[Steven Lord]

What you're trying to do sounds a lot like something that we strongly DISCOURAGE people from doing. See question 1 in the Programming section of the FAQ for more information.

[Image Analyst]

Well I thought of that at first too Steven, but then I thought of some cases where it might be used. For example the max() function can return the max, and optionally the index where the max occurs. So I suppose if you did not ask for the index, it wouldn't need to calculate it. Does it calculate it anyway even though it knows (at some point in time) that it should not return it?

[Walter Roberson]

Some routines calculate the extra variables anyhow and some do not. Mostly if the calculation of the variable is "expensive" then the calculation is skipped.

But that does not have to do with creation of variable names dynamically. Each potential output has a name, either named specifically on the left side of the "=" or as varargout{K} for some K. The variables are in the function workspace. When the function returns and there is an output argument to receive the value, everything except the name is copied over (and the reference count is fixed up if appropriate); if there is no slot to receive the value then it is discarded. Nothing dynamic about that.

[10B]

Thanks all for your comments. Steven, I hadn't come across that wiki (just searched through the Answers section for information), but I will refer there also as it seems to have a lot of useful info.

A colleague told me recently that a lot of mistakes I make are due to not knowing the 'rules' of programming. This is no doubt true and apparently why I asked my question - but it is hard to find the appropriate information on which rules to follow or how to apply them, when you don't know what the rules are in the first place!

I am working on an investigative project and I am not only trying to learn Matlab as I go, but programming as a whole due to simply not needing to use programming in my career previously. I have always been the end-user operating the GUI or similar.

Either way, thanks all for your thought and comments. I will no doubt ask more 'exploratory' questions in the future, and if you can forgive my errors, I will welcome your more experienced insights.

Answer 1

[Thorsten]

 temp = [10 2; 20 2; 30 1; 40 2; 50 1; 60 3];

 G = temp(temp(:,2) == 2, :);
 M = temp(temp(:,2) == 1, :);

If you like to do it in a function, define

 function Y = datagrab(X, ind)
 Y = X(X(:,2) == ind, :);

and call it using

 G = datagrab(temp, 2);
 M = datagrap(temp, 1);

[10B]

Thorsten - a really useful way to break it down for me so I can see where I went wrong with my earlier attempts. In fact, in comparison to what I tried this code is elegant in its simplicity! Thank you.

Answer 2

[Walter Roberson]

function datagrab(condition, varname)
  temp = .... some way of getting in the matrix

    find some matches and put their indexes into idx

    selected = temp(idx,:);
    assignin('caller', varname, selected)
  end

[10B]

Thank you for your code Walter - most useful. I can see how I can use this in future attempts.