Extrapolate the times of entry and exit in timetable

I have a timetable: I want to extrapolate the times of entry and exit from a certain room. I did a for loop on rows and two if loops: the first, I take the column for all the rooms on the i-th row (pir_tab_night.Column1streamId (i)) and I place it == at the room I'm interested in; I also added the condition that the start time must be 0 (in such a way as to take the first line in which the person enters the room) ... here it gives me error (Comparison is not defined between datetime and double arrays). the second if cycle is for the time out of the room: within this cycle I calculate the vect_duration for each (t_end-t_start) and then I go to concatenate with the function cat(1, vect_duration, t_end-t_start). Finally I repositioned the times = 0 to start a new count. This is my code:

vect_duration = []; t_start = 0; t_end = 0; for i = 1:length(pir_tab_night.t) if pir_tab_night.Column1streamId(i) == 'dbuid-17' & t_start == 0 t_start = pir_tab_night.t(i); end if pir_tab_night.Column1streamId(i) ~= 'dbuid-17' & t_start ~= 0 t_end = pir_tab_night.t(i-1); vect_duration = t_end - t_start; vect_duration = cat(1, vect_duration, t_end-t_start); t_start = 0; t_end = 0; end end

Can someone help me? Thank you

Answer 1

[Paolo]

 %Find dbuid-17 in your second column.
 dbuid17indx = pir_tab_night.Column1streamId==categorical({'dbuid-17'});

 %Use diff function to determine difference between entries.
 diffindx = diff(dbuid17indx);

 %Convert diffindx to char. Find patterns: 1, variable numbers of 0s, forward slash /.
 [startIndx,endIndx] = regexp(char(diffindx+'0')','(10*/)');

 %Group all entry and exit times.
 times = arrayfun(@(x,y) pir_tab_night.t(x+1:y),startIndx,endIndx,'un',0);

times is a cell array, where

 times{1} = 

   12-Sep-2017 21:38:45
   12-Sep-2017 21:39:33

 times{2} = 

   12-Sep-2017 23:23:43
   12-Sep-2017 23:24:19

 times{3} = 

   13-Sep-2017 04:37:12
   13-Sep-2017 04:39:04

 times{4} = 

   13-Sep-2017 06:26:34
   13-Sep-2017 06:26:55
   13-Sep-2017 06:27:32

... and so on. times contains the dates corresponding to every "entry" and "exit", as indicated by dbuid-17 in your second column.

[Erica Corradi]

Thanks for the reply! So the vector times contains all the input-output ranges including the values in between. it's right? If I wanted to calculate then the average time of all the input-output intervals (taking therefore only the start time and the end time of each) how should I do?

[Paolo]

That's correct. How do you wish to calculate the average time for when "times" only has one entry? In some of your entries for Column1streamId, sometimes dbuit-17 occurs only once.

[Erica Corradi]

yes, when there is only one row of dbuid-17, it means that the person has been in that room for 3 seconds

[Paolo]

 timespent = cellfun(@(x) x(end)-x(1),times);
 timespent(timespent==duration(0,0,0))=duration(0,0,3);
 avgtime = sum(timespent)/numel(timespent);
 avgtime = 

   00:09:56

[Erica Corradi]

Thank you so much for solving my problem !! Can I ask you to explain me why you put the +'0' and '(10*/)' in the regexp function?

[Peter Perkins]

This is a great vectorized solution to problem that seems like it would need a loop but really doesn't. A couple of minor readability suggestions:

1) it probably isn't necessary to explicitly construct a categorical value in creating dbuid17indx, I think this

dbuid17indx = pir_tab_night.Column1streamId=='dbuid-17';

should work.

2) Might be more readable to use seconds(3) rather than duration(0,0,3).

3) regexp is pretty challenging. If I'm reading the code correctly, it might be easier to just find positive and negative elements of diffindx.