How to count the number of non-nan values for combination of multiple variables

Hello, I would like to count the number of non-nan values in the d column for unique combinations of a, b and c (i.e I want to generate the e column in tt). If any a,b or c are NaN then the count should be nan as well

% Initial table
a = [1,1,1,2,2,2,2]';
b = [660, 661, 661, 663, 663, NaN, 663]';
c = [1,2,2,5,5,NaN,6]';
d = [11,12,NaN, 13, 14, NaN,5]';

t = table(a,b,c,d)

t = 7x4 table
    a     b      c      d 
    _    ___    ___    ___

    1    660      1     11
    1    661      2     12
    1    661      2    NaN
    2    663      5     13
    2    663      5     14
    2    NaN    NaN    NaN
    2    663      6      5

% Desired output
a = [1,1,1,2,2,2,2]';
b = [660, 661, 661, 663, 663, NaN, 663]';
c = [1,2,2,5,5,NaN,6]';
d = [11,12,NaN, 13, 14, NaN,5]';
e = [1, 1, 1, 2, 2, NaN, 1]';
tt = table(a,b,c,d,e)

tt = 7x5 table
    a     b      c      d      e 
    _    ___    ___    ___    ___

    1    660      1     11      1
    1    661      2     12      1
    1    661      2    NaN      1
    2    663      5     13      2
    2    663      5     14      2
    2    NaN    NaN    NaN    NaN
    2    663      6      5      1

Answer 1

[Voss]

% Initial table
a = [1;1;1;2;2;2;2];
b = [660; 661; 661; 663; 663; NaN; 663];
c = [1;2;2;5;5;NaN;6];
d = [11;12;NaN; 13; 14; NaN;5];

t = table(a,b,c,d)

t = 7x4 table
    a     b      c      d 
    _    ___    ___    ___

    1    660      1     11
    1    661      2     12
    1    661      2    NaN
    2    663      5     13
    2    663      5     14
    2    NaN    NaN    NaN
    2    663      6      5

G = {'a','b','c'};
F = @(x)nnz(~isnan(x));
T = groupsummary(t,G,F,'d');
[~,~,idx] = unique(t(:,G),'rows');
e = T.fun1_d(idx);
e(any(isnan(t{:,G}),2)) = NaN;
tt = addvars(t,e)

tt = 7x5 table
    a     b      c      d      e 
    _    ___    ___    ___    ___

    1    660      1     11      1
    1    661      2     12      1
    1    661      2    NaN      1
    2    663      5     13      2
    2    663      5     14      2
    2    NaN    NaN    NaN    NaN
    2    663      6      5      1

[Blue]

sweet :)

Answer 2

[J. Alex Lee]

Not sure what d and e need to be, but i think this should at least partially help

a = [1,1,1,2,2,2,2]';
b = [660, 661, 661, 663, 663, NaN, 663]';
c = [1,2,2,5,5,NaN,6]';

x = [a,b,c]

x = 7x3

1 660 1 1 661 2 1 661 2 2 663 5 2 663 5 2 NaN NaN 2 663 6

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

% any(_,2) goes row-wise ORs.
any(isnan(x),2)

ans = 7x1 logical array
   0
   0
   0
   0
   0
   1
   0