Keeping the same order in arrays - Brain Teaser?

**EDIT: I need the output to be a matrix because a much larger part of the code needs the output of this code to be a matrix for it's input****

Hello. I have a question related to matrix manipulation.

I need to keep the order the same in the matrix. Please see below for what I am trying to do.

lets say I start out with 3 fruits:

    fruits = {'apple','orange','berry'};

and the amounts of each fruit:

    amount = [3,5,2]

then the next day the amount changes:

    amount = [2,4,3]

so now, my matrix will be:

    3  5  2
    2  4  3

but what if the next day I needed to add another fruit:

   fruits = {'apple','orange','berry','banana'};

and the amounts are:

   amount = [3,4,2,1]

how do I make my new matrix like this:

   3  5  2  NaN
   2  4  3  NaN
   3  4  2  1

then on the next day, I was not given 1 of the original fruits:

   fruits = {'apple','berry','banana'};

and the amount would be:

   amount = [5,1,4]

then I need the matrix to be like the following:

   3  5  2  NaN
   2  4  3  NaN
   3  4  2  1
   5  NaN  1  4

How would I write the code for it to be able to handle all of these situations?

[Ingrid]

is there any particular reason why you would want this in matrix format?

are you aware of the possibility to use structures in matlab with dynamic indexing? than you would do something like this for each time t:

for ii = 1:numel(field(fruits))
if isfield(countingFruits.(fruits)
         countingFruits.(fruits{ii}) = [countingFruits.(fruits{ii}); t, amount];
else
         countingFruits.(fruits{ii}) = [t,amount(ii)];
end
end

you need to save the t to later be able to construct NaN for missing t's

[Viral Solanki]

Yes, I did think of many other ways of doing this, but it needs to be in this format because of other integrated code. The other code I'm working with needs this particular input

[Viral Solanki]

I also could do it with a bunch of ifs and loops, but I wanted to learn about how to use matrix manipulation techniques. Don't want to keep making new dirty code

Answer 1

[Stephen23]

Try using a non-scalar structure (note that looping backwards is useful and not a mistake)

fruits{4} = {'apple','berry','banana'};
amount{4} = [5,1,4];
fruits{3} = {'apple','orange','berry','banana'};
amount{3} = [3,4,2,1];
fruits{2} = {'apple','orange','berry'};
amount{2} = [2,4,3];
fruits{1} = {'apple','orange','berry'};
amount{1} = [3,5,2];
for mm = numel(fruits):-1:1 % backwards!
    for nn = 1:numel(fruits{mm})
        out(mm).(fruits{mm}{nn}) = amount{mm}(nn);
    end
end

And we can view the data:

>> {out.apple}
ans = 
    [3]    [2]    [3]    [5]
>> {out.banana}
ans = 
    []    []    [1]    [4]

Note the missing data are indicated by empty arrays.

[Viral Solanki]

This is a great structure, but unfortuntely I need the output to be in matrix format because of the input into another part of a much bigger code. Thank you so much

[Stephen23]

Add these three lines to turn that structure into a numeric array:

tmp = squeeze(struct2cell(out));
tmp(cellfun('isempty',tmp)) = {NaN};
cell2mat(tmp')

And here it is displayed in the command window:

ans =
   3     2   NaN     5
   2     3   NaN     4
   3     2     1     4
   5     1     4   NaN

You can get the list of fruit corresponding to each column:

>> fieldnames(out)
ans = 
    'apple'
    'orange'
    'berry'
    'banana'

[Viral Solanki]

you are THE man! thank you so much.

[Viral Solanki]

I can not even explain how grateful I am to you!

Answer 2

[Steven Lord]

Try using a table.

t = array2table([3 5 2], 'VariableNames', {'apple','orange','berry'});
t(end+1, {'apple', 'orange', 'berry'}) = {2, 4, 3};
t(end+1, {'apple', 'orange', 'berry', 'banana'}) = {3, 4, 2, 1};
t(end+1, {'apple', 'berry', 'banana'}) = {5, 1, 4};

This will give you a warning on the last line, but if you intentionally did not assign to the orange variable you can suppress it. It will also fill in with 0's instead of NaN but you can use standardizeMissing to replace 0 with NaN. See the examples in the table documentation for more information on how to work with this data type.

[Viral Solanki]

This would be awesome, but there are 2 issues.

1. I am using an older version of matlab (R2012B) and

2. The code that is already integrated into the system needs a matrix as an input.

But, thank you for this. I will remember this