How to convert dates and times from txt to numbers?

I have two txt files! The first one contains these data. It displays the activities of a person for different times.

2011-11-28 02:27:59    2011-11-28 10:18:11    Sleeping    
2011-11-28 10:21:24    2011-11-28 10:23:36    Toileting  
2011-11-28 10:25:44    2011-11-28 10:33:00    Showering  
2011-11-28 10:34:23    2011-11-28 10:43:00    Breakfast  
2011-11-28 10:49:48    2011-11-28 10:51:13    Grooming  
2011-11-28 10:51:41    2011-11-28 13:05:07    Spare_Time/TV    
2011-11-28 13:06:04    2011-11-28 13:06:31    Toileting  
2011-11-28 13:09:31    2011-11-28 13:29:09    Leaving

This is my second txt file.This txt displays the time where a sensor is open for different times.. I want to read the whole txts and convert the dates and times from each row to numbers, in order to compare them in the end. I assume the activities with these kind of settings: Leaving --> 1 Toileting --> 2 Showering --> 3 Sleeping --> 4 Breakfast --> 5 Lunch --> 6 Dinner --> 7 Snack --> 8 Spare_Time/TV --> 9 Grooming --> 10. So my array would look like this: target(p,:) = [0, 0, 0, 1, 0, 0, 0, 0, 0, 0]. I put 1, because the sensor(bed) is open for the activity sleeping the curent time and 0 where the sensor is not open.

Start time            End time              Location  Type    Place
--------------------  --------------------  --------  --------  -----
2011-11-28 02:27:59    2011-11-28 10:18:11    Bed      Pressure  Bedroom
2011-11-28 10:21:24    2011-11-28 10:21:31    Cabinet    Magnetic  Bathroom
2011-11-28 10:21:44    2011-11-28 10:23:31    Basin    PIR      Bathroom
2011-11-28 10:23:02    2011-11-28 10:23:36    Toilet    Flush    Bathroom
2011-11-28 10:25:44    2011-11-28 10:32:06    Shower    PIR      Bathroom
2011-11-28 10:34:23    2011-11-28 10:34:41    Fridge    Magnetic  Kitchen
2011-11-28 10:34:44    2011-11-28 10:37:17    Cupboard  Magnetic  Kitchen
2011-11-28 10:38:00    2011-11-28 10:42:41    Toaster    Electric  Kitchen
2011-11-28 10:38:33    2011-11-28 10:38:40    Fridge    Magnetic  Kitchen
2011-11-28 10:41:29    2011-11-28 10:41:36    Cupboard  Magnetic  Kitchen
2011-11-28 10:41:43    2011-11-28 10:41:59    Cooktop    PIR      Kitchen
2011-11-28 10:41:59    2011-11-28 10:42:55    Microwave  Electric  Kitchen

The sensors should be like this: Shower --> 1 Basin --> 2 Cooktop --> 3 Maindoor --> 4 Fridge --> 5 Cabinet --> 6 Cupboard --> 7 Toilet --> 8 Seat --> 9 Bed --> 10 Microwave --> 11 Toaster --> 12

This array should look like this: x(p,:) = [0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0], because sensor bed is open this current time. I tried to use a different method, but i doesn't seem to work properly. Below is my code,but i want to use the above method.

clear; clc; close all;
fh = fopen('OrdonezA_ADLs.txt','r');
Data=textscan(fh, '%s%s%s%s%s');
fclose(fh);
n=size(Data,1);       
n = n-4;             
P = length(Data{1});  

x = zeros(n,P); 
for pat=1:P
    for i=1:n
        x(i,pat) = double(Data{i+4}(pat));
    end
end

t = zeros(1,P); % targets
for pat=1:P
    if strcmp(Data{1,5}(pat),'Sleeping')  
        t(pat) = 1;

      elseif strcmp(Data{1,5}(pat),'Toileting')  
          t(pat) = 1;

      elseif strcmp(Data{1,5}(pat),'Spare_Time/TV')  
          t(pat) = 1;

      elseif strcmp(Data{1,5}(pat),'Grooming')  
          t(pat) = 1;

      elseif strcmp(Data{1,5}(pat),'Snack')  
          t(pat) = 1;

      elseif strcmp(Data{1,5}(pat),'Showering')  
          t(pat) = 1;     
      else
          t(pat) = 0;

      end

end

[dpb]

"Sensors: Shower--> 1 Basin --> 2 Cooktop --> 3 ... Toaster --> 12"

>> locs={'Shower' 'Basin' 'Cooktop' 'Maindoor' 'Fridge'  'Cabinet' 'Cupboard' 'Toilet' 'Seat' 'Bed' 'Microwave' 'Toaster'}.';
>> idx=[1:length(sensorlocs)].';
>> sensors=categorical(idx,idx,locs)
sensors = 
   Shower 
   Basin 
   Cooktop 
   Maindoor 
   Fridge 
   Cabinet 
   Cupboard 
   Toilet 
   Seat 
   Bed 
   Microwave 
   Toaster 
>> double(sensors)
ans =
   1
   2
   3
   4
   5
   6
   7
   8
   9
  10
  11
  12
>>

Best of both worlds--the variables are identified by something meaningful but the values can still be 1:N if you think they must or the outside data comes in as numeric instead of the string representation. But, that isn't what your data file indicates, it's using the string so really would be easier to use it as is. It doesn't really appear there's any need for the ordinal sequence number other than that's just what you started with...what does the number matter if you can keep track of them by name instead?

Answer 1

[dpb]

I'd suggest using table for such data/analyses. I put your first file into a local file 'kpl.txt' and manipulated it with a few simple machinations to achieve--

>> kpl=readtable('kpl.txt','delimiter','\t','readvariablenames',0);  % read the data
>> kpl=kpl(:,~all(ismissing(kpl)));                                  % eliminate empty from multiple delimiters
>> kpl.Properties.VariableNames={'Start' 'End' 'Activity'}           % meaningful column names
kpl = 
          Start                     End                Activity    
  _____________________    _____________________    _______________
  '2011-11-28 02:27:59'    '2011-11-28 10:18:11'    'Sleeping'     
  '2011-11-28 10:21:24'    '2011-11-28 10:23:36'    'Toileting'    
  '2011-11-28 10:25:44'    '2011-11-28 10:33:00'    'Showering'    
  '2011-11-28 10:34:23'    '2011-11-28 10:43:00'    'Breakfast'    
  '2011-11-28 10:49:48'    '2011-11-28 10:51:13'    'Grooming'     
  '2011-11-28 10:51:41'    '2011-11-28 13:05:07'    'Spare_Time/TV'
  '2011-11-28 13:06:04'    '2011-11-28 13:06:31'    'Toileting'    
  '2011-11-28 13:09:31'    '2011-11-28 13:29:09'    'Leaving'      
>>

At this point all are cell arrays; let's turn into useful forms of datetime and categorical for the activities--

>> kpl.Start=datetime(kpl.Start,'inputformat','yyyy-MM-dd HH:mm:ss');
>> kpl.End=datetime(kpl.End,'inputformat','yyyy-MM-dd HH:mm:ss');
>> kpl.Activity=categorical(kpl.Activity);
>> kpl.Duration=(kpl.End-kpl.Start)        % also would like the time each takes
kpl = 
         Start                    End               Activity       Duration
  ____________________    ____________________    _____________    ________
  28-Nov-2011 02:27:59    28-Nov-2011 10:18:11    Sleeping         07:50:12
  28-Nov-2011 10:21:24    28-Nov-2011 10:23:36    Toileting        00:02:12
  28-Nov-2011 10:25:44    28-Nov-2011 10:33:00    Showering        00:07:16
  28-Nov-2011 10:34:23    28-Nov-2011 10:43:00    Breakfast        00:08:37
  28-Nov-2011 10:49:48    28-Nov-2011 10:51:13    Grooming         00:01:25
  28-Nov-2011 10:51:41    28-Nov-2011 13:05:07    Spare_Time/TV    02:13:26
  28-Nov-2011 13:06:04    28-Nov-2011 13:06:31    Toileting        00:00:27
  28-Nov-2011 13:09:31    28-Nov-2011 13:29:09    Leaving          00:19:38
>>

At this point you've got the above table to query for any activity or interval as desired.

>> varfun(@sum,kpl,'groupingvariable','Activity','inputvariables','Duration')
ans = 
                     Activity       GroupCount    sum_Duration
                   _____________    __________    ____________
  Breakfast        Breakfast        1             00:08:37    
  Grooming         Grooming         1             00:01:25    
  Leaving          Leaving          1             00:19:38    
  Showering        Showering        1             00:07:16    
  Sleeping         Sleeping         1             07:50:12    
  Spare_Time/TV    Spare_Time/TV    1             02:13:26    
  Toileting        Toileting        2             00:02:39    
>>

Only slight modifications would be needed for the other file to create corollary table from it.

[kpl]

I used your code,but i get this error message:

Error using readtable (line 143)
Reading failed at line 15. *All lines of a text file must have the same number of delimiters*. Line 15
has 9 delimiters, while preceding lines have 7.

This is how i wrote it

kpl=readtable('OrdonezA_ADLs.txt','delimiter','\t','readvariablenames',0);  % read the data
kpl=kpl(:,~all(ismissing(kpl)));
kpl.Properties.VariableNames={'Start' 'End' 'Activity'};
kpl.Start=datetime(kpl.Start,'inputformat','yyyy-MM-dd HH:mm:ss');
kpl.End=datetime(kpl.End,'inputformat','yyyy-MM-dd HH:mm:ss');
kpl.Activity=categorical(kpl.Activity);
kpl.Duration=(kpl.End-kpl.Start);

kpl2=readtable('OrdonezA_Sensors.txt','delimiter','\t','readvariablenames',0);  % read the data
kpl2=kpl2(:,~all(ismissing(kpl2)));
kpl2.Properties.VariableNames={'Start' 'End' 'Activity'};
kpl2.Start=datetime(kpl2.Start,'inputformat','yyyy-MM-dd HH:mm:ss');
kpl2.End=datetime(kpl2.End,'inputformat','yyyy-MM-dd HH:mm:ss');
kpl2.Activity=categorical(kpl2.Activity);
kpl2.Duration=(kpl2.End-kpl2.Start);

[dpb]

It tells you exactly what and where the problem is--

"Reading failed at line 15. All lines of a text file must have the same number of delimiters. Line 15 has 9 delimiters, while preceding lines have 7."

Fix the error in the input file. I noted above there were multiple delimiters, but all lines were the same at least. Your file isn't even that consistent.

Matlab is an excellent product, but it can't debug/correct the input files for you.

[kpl]

The txt files are so bad written.Although i managed to read them.The code about txt files is now working.In my second txt file, i get a plus variable,but i don't know why.Here are the changes i've done:

kpl=readtable('OrdonezA_ADLs.txt','delimiter','\t','readvariablenames',0);  % read the data
kpl=kpl(:,~all(ismissing(kpl)));
kpl.Properties.VariableNames={'Start' 'End' 'Activity'};

P = length(kpl{1,:});
n=size(kpl,1); 

x = zeros(n,P); 
for pat=1:P
    for i=1:n
        x(i,pat) =(kpl{i,:}(pat));
    end
end

kpl.Start=datetime(kpl.Start,'inputformat','yyyy-MM-dd HH:mm:ss');
kpl.End=datetime(kpl.End,'inputformat','yyyy-MM-dd HH:mm:ss');
kpl.Activity=categorical(kpl.Activity);
kpl.Duration=(kpl.End-kpl.Start);

kpl2=readtable('OrdonezA_Sensors.txt','delimiter','\t','readvariablenames',0);  % read the data
kpl2=kpl2(:,~all(ismissing(kpl2)));
kpl2.Properties.VariableNames={'Start' 'vf' 'End' 'Activity'};
kpl2.Start=datetime(kpl2.Start,'inputformat','yyyy-MM-dd HH:mm:ss');
kpl2.End=datetime(kpl2.End,'inputformat','yyyy-MM-dd HH:mm:ss');
kpl2.Activity=categorical(kpl2.Activity);
kpl2.Duration=(kpl2.End-kpl2.Start);

Since the code almost completely changed, it gives me new error message in this line, saying that ' Conversion to double from cell is not possible.' How do i fix this?

x(i,pat) =(kpl{i,:}(pat));.

[dpb]

Dunno...what's all that for, anyway? As is, however, you've tried to mush different types of variables altogether in an array which, as you've discovered, doesn't work. But, the important point is there's no need for it--we went to all the trouble to build the table, now use the facilities of the table, don't just throw it away or leave it gathering dust in the corner.

Yes, this may mean you need to read the documentation and see how its used in some of the examples... :)

The whole point of the table as illustrated for the first file is that you can do all the computations you're trying to do with these complex machinations and loops and string searches inside the table with categorical and datetime/duration variables in just one or two lines each.

PS: I'd recommend using a more meaningful name for the table variables; I just used your user name as a temporary here to illustrate the concepts; in real code having the names reflect something about the content would be "much more better"...

[dpb]

So, let's step back a bit and instead of looking at the code you've attempted to write, explain what you want to get from the two files. You've got durations for activities and another set for something about a physical item but the two don't coincide so can't join the two.

You're trying to build some indicator array, but it's not clear what the actual meaning is to be and what it is you're trying to get to as the end result.

The question of setting some flag based on activity as compared to what article is in use is easily enough done through lookup altho you need a 1:1 or 1:many lookup table stored to make the connection; what you've got hardocded into the initial loop with string matching but just as a table for lookup/association.

But, a more high-level objective of what you want to retrieve from these files would serve well to allow for an efficient solution.

[kpl]

I want to put the first file inside an array on matlab in order to perform cross validation method with rbf kernel. n is the number of features and p is the number of patterns. I need to read those two files, to compare the times and put 1 where the sensor is open and 0 where the sensor is close and put those results in a table t.

[dpb]

OK, but I'm still somewhat confused...compare which times, precisely? That is, what times are to be those in the output array since what is in the two files are durations, not specific times.

Seems to me you need a set of actual point times at which to take snapshots of activities...or how do you propose to do this comparison, exactly? "Needs to know" the "which time?" question to do the comparison/selection.

[kpl]

The times of the first file on each row. eg

2011-11-28 02:27:59'    '2011-11-28 10:18:11'    'Sleeping'     

with the times of the second file eg. 

      2011-11-28 02:27:59    2011-11-28 10:18:11    Bed

Compare each files time and if they do match or they are inside the time limit then put 1 on t table(1 open sensor) and if they dont match put 0 on table t(close sensor). In this example times match, so i put 1 on t table.For the next rows times maybe not match between them.

[dpb]

Hmmm....that just doesn't seem exactly what I'd think you'd want, but you're the boss... :)

OK, first get the time overlap...that's pretty simple; let's define an anonymous function

fnOverlap=@(ts,te,Ts,Te) (te>=Ts) & (ts<=Te);  % t interval inside interval T

to return a logical value T when the time intervals overlap. Then the intersection of the number of intervals is number rows in T1 by rows in T2.

timtab=false(height(A),height(B);  % just use A, B for table names
for for i=1:height(kpl)
  timtab(i,:)=fnOverlap(A{i,'Start'},A{i,'End'},B.Start,B.End).';
end

For the two here I get

>> timtab
timtab =
   1     0     0     0     0     0     0     0     0     0     0     0
   0     1     1     1     0     0     0     0     0     0     0     0
   0     0     0     0     1     0     0     0     0     0     0     0
   0     0     0     0     0     1     1     1     1     1     1     1
   0     0     0     0     0     0     0     0     0     0     0     0
   0     0     0     0     0     0     0     0     0     0     0     0
   0     0     0     0     0     0     0     0     0     0     0     0
   0     0     0     0     0     0     0     0     0     0     0     0
>>

which is a row for each time in A and which interval in B is contained if (if any).

Ewww....I've got a noon commitment...the step to pick the proper sensor for the overlapping time shouldn't be too much of a stretch.

[dpb]

PS: A quick glance looked like the logic above worked correctly; you should verify if I did overlook something; I didn't make it exhaustive check.

[kpl]

It works.The only thing that matters now is table x. I want to fill x with all the values from the first file like this:

P = length(kpl{1,:});
n = size(kpl,1); 

x = zeros(n,P); 
for pat=1:P
    for i=1:n
        x(i,pat) =(kpl{i,:}(pat));
    end
end

Although, i get an error on this line

x(i,pat) =(kpl{i,:}(pat));

I don't want to use this command

x = readtable('OrdonezA_ADLs.txt');

[dpb]

I can't even get the first line to work to determine P. We built the tables, let's use them; they're much more convenient than trying to remember indices and stuff.

Answer 2

[dpb]

OK, with the updated information on what is wanted the overall code is accumulated here in one spot instead of in the string of comments piecemeal--

First, I'll presume you can clean up the data files to eliminate the duplicated delimiters and so on--that's outside the domain of solving the actual problem. There's only one detail that was omitted in creating the tables initially--maintaining the desired order of categorical variables as you're designated them to be; initially I really didn't think it would matter and it really doesn't other than you do need a correspondence so we can use the one you've already defined--

sensorlocs={'Shower' 'Basin' 'Cooktop' 'Maindoor' 'Fridge' 'Cabinet' 'Cupboard' 'Toilet' 'Seat' 'Bed' 'Microwave' 'Toaster'}.';
activities={'Leaving' 'Toileting' 'Showering' 'Sleeping' 'Breakfast' 'Lunch' 'Dinner' 'Snack' 'Spare_Time/TV' 'Grooming'}.';
act=readtable('kpl.txt','delimiter','\t','readvariablenames',0);  % read the activity data
act=act(:,~all(ismissing(act)));                                  % eliminate empty rows
act.Properties.VariableNames={'Start' 'End' 'Activity'};          % meaningful column names
act.Start=datetime(act.Start,'inputformat','yyyy-MM-dd HH:mm:ss');
act.End=datetime(act.End,'inputformat','yyyy-MM-dd HH:mm:ss');
act.Activity=categorical(act.Activity,activities);                % use assigned numerical order
act.Duration=(act.End-kpl.Start);

I cleaned up the other dataset(*); with that I could simply

sens=readtable('OrdonezA_Sensors.txt');
sens.Location=categorical(sens.Location,sensorlocs);
sens.Type=categorical(sens.Type);
sens.Place=categorical(sens.Place);

and end up with the usable table. If there are specific orders for Location and Place, define and use them analogously as to Location; you didn't give those correlations. From this morning for the time table--

fnOverlap=@(ts,te,Ts,Te) (te>=Ts) & (ts<=Te);  % t interval inside interval T
timtab=false(height(act),height(sens);
for i=1:height(act)
  timtab(i,:)=fnOverlap(act{i,'Start'},act{i,'End'},sens.Start,sens.End).';
end
% compute the sensor truth table from the time sequencing
senstab=false(height(act),height(sens);
for i=1:height(act)
  senstab(i,sens{timtab(i,:),'Location'))=true;
end

The result for the two datasets provided looks like:

>> senstab
senstab =
   0     0     0     0     0     0     0     0     0     1     0     0
   0     1     0     0     0     1     0     1     0     0     0     0
   1     0     0     0     0     0     0     0     0     0     0     0
   0     0     1     0     1     0     1     0     0     0     1     1
   0     0     0     0     0     0     0     0     0     0     0     0
   0     0     0     0     0     0     0     0     0     0     0     0
   0     0     0     0     0     0     0     0     0     0     0     0
   0     0     0     0     0     0     0     0     0     0     0     0
>>

NB: The two loops both operate over the first table rows and the locations of interest are found in the output from the anonymous function so there's really no reason one couldn't merge the two and even do away with the time overlap array entirely altho I figured it would likely be of interest somewhere on down the road.

Anyway, that seems to me far simpler than trying to deal with cell arrays and combinations of different data types and such explicitly--the table is really suited to handle such disparate types compactly and with highly readable (and hence maintainable) code plus expanding it to handle more sensors or locations or whatever is essentially trivial. Plus, the other analysis functions can be brought to bear to do much more in-depth probing of the data.

Granted, it will take a little bit of reading up on the table data class, but think the investment will save much more time going forward than the initial effort will take.

(*) See followup comment

[dpb]

(*) The cleaned-up file looks like the following. If you can't get the originator of the files to clean them up when they're being generated, I'd suggest writing a filter to do so independently of trying to use them.

Start,End,Location,Type,Place
2011-11-28 02:27:59, 2011-11-28 10:18:11,  Bed,       Pressure, Bedroom
2011-11-28 10:21:24, 2011-11-28 10:21:31,  Cabinet,   Magnetic, Bathroom
2011-11-28 10:21:44, 2011-11-28 10:23:31,  Basin,     PIR,      Bathroom
2011-11-28 10:23:02, 2011-11-28 10:23:36,  Toilet,    Flush,    Bathroom
2011-11-28 10:25:44, 2011-11-28 10:32:06,  Shower,    PIR,      Bathroom
2011-11-28 10:34:23, 2011-11-28 10:34:41,  Fridge,    Magnetic, Kitchen
2011-11-28 10:34:44, 2011-11-28 10:37:17,  Cupboard,  Magnetic, Kitchen
2011-11-28 10:38:00, 2011-11-28 10:42:41,  Toaster,   Electric, Kitchen
2011-11-28 10:38:33, 2011-11-28 10:38:40,  Fridge,    Magnetic, Kitchen
2011-11-28 10:41:29, 2011-11-28 10:41:36,  Cupboard,  Magnetic, Kitchen
2011-11-28 10:41:43, 2011-11-28 10:41:59,  Cooktop,   PIR,      Kitchen
2011-11-28 10:41:59, 2011-11-28 10:42:55,  Microwave, Electric, Kitchen

[dpb]

BTW, if you haven't yet concluded that the table and categorical variables make referencing, etc., worth the effort, let me show another result from having used same on the above table -- it's pretty hard to check things are correct just from inspection, right? How about

>> [i j]=find(tabsens.');        % find locations by column,row
>> c=categories(sens.Location);  % get the value names in array
% and display by row the locations and corresponding location
>> disp([num2str([j i],'%4d %4d  ') char(arrayfun(@(i) c{i},i,'uniformoutput',0))])
1   10  Bed      
2    2  Basin    
2    6  Cabinet  
2    8  Toilet   
3    1  Shower   
4    3  Cooktop  
4    5  Fridge   
4    7  Cupboard 
4   11  Microwave
4   12  Toaster  
>>

Isn't having the name to look at a lot easier to debug or report than just the number '8', say??? Who knows what "8" is and it'll be something different for another variable, besides.

[kpl]

The only code i managed to get it το work is this:

% Read Daily Living Data
clear; clc; close all;
% Read Daily Living Data
clear; clc; close all;

fh = fopen('OrdonezA_ADLs.txt','r');
Data=textscan(fh, '%s%s%s%s%s');
fclose(fh);

n=size(Data,1);       
n = n-4;             
P = length(Data{1});  

x = zeros(n,P); 
for pat=1:P
    for i=1:n
        x(i,pat) = double(Data{i+4}(pat));
    end
end

Data=readtable('OrdonezA_ADLs.txt','delimiter','\t','readvariablenames',0);  % read the data
Data=Data(:,~all(ismissing(Data)));
Data.Properties.VariableNames={'Start' 'End' 'Activity'};
Data.Start=datetime(Data.Start,'inputformat','yyyy-MM-dd HH:mm:ss');
Data.End=datetime(Data.End,'inputformat','yyyy-MM-dd HH:mm:ss');
Data.Activity=categorical(Data.Activity);
Data.Duration=(Data.End-Data.Start);

kpl2=readtable('OrdonezA_Sensors.txt','delimiter','\t','readvariablenames',0);  % read the data
kpl2=kpl2(:,~all(ismissing(kpl2)));
kpl2.Properties.VariableNames={'Start' 'vf' 'End' 'Activity'};
kpl2.Start=datetime(kpl2.Start,'inputformat','yyyy-MM-dd HH:mm:ss');
kpl2.End=datetime(kpl2.End,'inputformat','yyyy-MM-dd HH:mm:ss');
kpl2.Activity=categorical(kpl2.Activity);
kpl2.Duration=(kpl2.End-kpl2.Start);

kpl2(:,2) = [];

t = zeros(1,P); % targets
for pat=1:P
    if strcmp(Data.Duration,kpl2.Duration)

      t(pat)=1;

      else
          t(pat)=0;

      end

end

I use strcmp to compare the duration times for each line from both files, but the line does not work. Any ideas how to fix this?

[dpb]

Yes, use the datetime and duration data as their actual data types--they're NOT strings(); that's just the visual representation for display. Similarly like the |*categorical| data values internally are integer.

Besides that, even if were strings strcmp is an exact match so comparing the durations would only return T for the one case (the first record in each dataset); there are no other identical durations.

If there were a reason to compare durations, however, it's simply a numeric comparison...

>> kpl.Duration(1)==sens.Duration.'  % compare first to table 2
ans =
   1     0     0     0     0     0     0     0     0     0     0     0
>>

This of course shows that there is only that one record; the rest of a loop over the first table comparing to the second would all be F (0). That's just not the lookup you're after; it's the intersection between start-end of one lying within the same interval of the other that is the test I wrote in the anonymous function for convenience in its application.

What's the problem with using the code I provided or other similar to it? It's why I provided it; to give you an outline of how to effectively process data such as you have. Feel free to modify it as wanted/needed, of course, but the fundamental process of how to manipulate datetimes and categorical-type variables is key.

Q: Is not senstab shown above precisely the table you were trying to create in the request "Compare each file's time and if they do match or they are inside the time limit then put 1 on t table..."? As the followup above shows, the sensors match the positions in the second file table at the time interval(s) of the first file.

() That was the point of the |*datetime| conversion in response to the question subject--"turn date strings into numbers?".

[dpb]

See the sections under the link at <DATETIME Operations>, particularly the two on date arithmetic and comparisons; it'll explain the concepts and show examples and you'll continue to beat your head against a dead end alley ignoring the concepts therein.

Similarly, read and study the sections under <Data Types> particularly the categorical and table types. With R2016b there's also a timetable type, but even if you have that late a release it is designed for a specific time per record, not intervals such as your data so don't think it'll help much; the table really is the best structure here imo, you need to allow it to help instead of fighting against it.

[dpb]

One last effort...

"The only code i managed to get it το work is this:"

fh = fopen('OrdonezA_ADLs.txt','r');
Data=textscan(fh, '%s%s%s%s%s');
fclose(fh);

There's no point in the above; readtable following reads the data and we subsequently massage it into the needed forms of datetime for the dates and categorical for the other variables of locations, activities, etc., etc., that they are that gives the best of both worlds as well as does the datetime class--a numeric representation internally that allows for comparisons and tests for equality easily plus the ability to actual have human-readable outputs for reporting and debugging.

Plus, it's simple to build the "regular" array of "when who does what and where" by just straight logical addressing making use of the fact the categorical arrays ARE numeric, their values are then the indices into the array directly.

So, as noted before, once you do read the data into the table, there's no reason at all to revert to trying to access cell arrays of strings with all the perils of only having numeric column indexing to keep track of--the table gives you meaningful names to use for each as well as for the values.

 "Use the tools available at hand, Luke!!!" :)