Convert digital numbers to datetime elements

I have two types of time arrays, t1 and t2, with the same number of elements, i.e. 1056, and both indicating a time range of 44 days.

The number 1056 comes from 24 hours (in a day) times 44 days, which represents, indeed, 1056 hours.

The t1 array contains decimal numbers between 0 and 44, while the t2 array contains datetime elements, as shown here below:

% array t1
t1 = linspace(0,44,1057);

>> t1

t1 =
1056×1 single column vector

0
0.04166667
0.08333334
0.125
0.1666667
0.2083333
0.25
0.2916667
0.3333333
...
43.83331
43.87498
43.91665
43.95831

% array t2
a = {'17-Jun-2017'; '30-Jul-2017'};
b = cellstr(datetime(a{1}) : days(1) : datetime(a{end}))
k = 1;
for i = 1:length(b)
    for j = 0:23
        t2{k} = strjoin([b(i),sprintf('%d:00:00',j)]);
        k = k + 1;
    end
end

>> t2'

ans =
1056×1 cell array

{'17-Jun-2017 0:00:00' }
{'17-Jun-2017 1:00:00' }
{'17-Jun-2017 2:00:00' }
{'17-Jun-2017 3:00:00' }
{'17-Jun-2017 4:00:00' }
{'17-Jun-2017 5:00:00' }
...
{'30-Jul-2017 19:00:00'}
{'30-Jul-2017 20:00:00'}
{'30-Jul-2017 21:00:00'}
{'30-Jul-2017 22:00:00'}
{'30-Jul-2017 23:00:00'}

Now, if I pick up a random decimal number between 0 and 44, but not necessarily one of those ones contained in the t1 array, how can I convert that decimal number into a datetime element, which would correspond to something between the elements of t2 ?

I know, my question is a bit unclear, but let me give you an example. We have the t1 and t2 arrays:

t1                                  t2
--------------------------------------------------
0          corresponds to {'17-Jun-2017 0:00:00' }
0.04166667 corresponds to {'17-Jun-2017 1:00:00' }
0.08333334 corresponds to {'17-Jun-2017 2:00:00' }
0.125      corresponds to {'17-Jun-2017 3:00:00' }
...        corresponds to ...
--------------------------------------------------

If I pick up

0.1 

I guess it will correspond to something similar to:

{'17-Jun-2017 2:30:00' }

Pay attention to the time, which is 2:30 in this example, i.e. between 2:00 and 3:00, both elements of the t2 array.

Therefore, how can I get that

{'17-Jun-2017 2:30:00' }

?

Any magical matlab-trick? :-)

Answer 1

[Voss]

If you make a datetime array from t2, you can use interp1 to interpolate.

(Also, t1 has 1057 elements, so I'll remove the last one, which corresponds to midnight at the end of 30-Jul-2017, which does not appear in t2.)

% array t1
t1 = linspace(0,44,1057);

% remove the last element of t1:
t1(end) = [];

% array t2
a = {'17-Jun-2017'; '30-Jul-2017'};
b = cellstr(datetime(a{1}) : days(1) : datetime(a{end}));
k = 1;
for i = 1:length(b)
    for j = 0:23
        t2{k} = strjoin([b(i),sprintf('%d:00:00',j)]);
        k = k + 1;
    end
end

% pick 0.1:
t_pick = 0.1;

% interpolate:
interp1(t1(:),datetime(t2(:)),t_pick)

ans = datetime
   17-Jun-2017 02:24:00

You say t2 is a datetime array, but actually it's a cell array of character vectors. If you are constructing it yourself as shown, you're probably better off making it a datetime array.

[Sim]

Thanks a lot @_ !! Yes, the interp1 function, I did not think about it at all.... :-)

It works!

And thank you for correcting my (incorrect) terminology, that I have used in this post!

:-)

A small note... If the t1 array is a "single" type array, i.e.

>> class(t1)
ans =
    'single'

the interpolation function does not work , giving me an error.

As workaround, I have just converted t1, from a "single" type to a "double" type one:

t1 = double(t1)

and after that, the interpolation function

interp1(t1(:),datetime(t2(:)),t_pick)

worked again :-)

[Voss]

You're welcome!

Good catch, on the problem with using interp1 with single and datetime inputs. I think converting t1 to double is the way to go.

[Sim]

thanks a lot for your comment :)

[Voss]

You're welcome!