MATLAB: Retime only for specific gaps on time

retime

I have a timetable z with more than 10 000 rows, this timetable should have a resolution of 15 minutes, but there are gaps of for example 30,45 minutes, 1, 5, 8 hours, or even days. So I would like to fill the missing data using retime but only for the gaps lower or equal to a specific gap (for example 1 hour). How can I do that? I have search for ways of doing that, but have found no hint. I am using this line, but it fills all the gaps.

retime(z,'regular','linear','TimeStep',minutes(15))

Any help would be very much appreciatted.

Best Answer

How about the following solution?

% Sample timetable with 2 gaps (e.g >1 hour)
Time = datetime('now') + minutes(cumsum(45*rand(100,1)));
Time(30:end) = Time(30:end) + hours(1); % Gap1
Time(60:end) = Time(60:end) + hours(1); % Gap2
Data = rand(100,1);
TT = timetable(Time,Data);
% Find gaps and create group vector
idx = diff(TT.Time) > hours(1);
idx = [true; idx];
group = cumsum(idx);
% Apply retime function for each group and concatenate the result
TT2 = [];
for kk = 1:max(group)
  idx = group == kk;
  TT2 = [TT2; retime(TT(idx,:),'regular','linear','TimeStep',minutes(15))];
end

Related Solutions

MATLAB: Sequence of date with no leap year

Add this line after you create your Time variable:

Time(month(Time)==2 & day(Time)==29) = []

MATLAB: I have a time table with a year’s worth of current measurements. (Time format is Year-Month-Day-Hour-Min-Sec) How to take each day and count how many times the current was over a specific reading that day.

The method input argument to retime can be a function handle to a function that accepts a vector of data and returns a scalar or row vector. Try writing a function that accepts the data from each day and returns the number of values that are greater than your threshold and passing that into retime as the method input.

Let's start with some random data.

rng default
mpd = minutes(days(1)); % minutes per day
T = datetime('today');
x = randi(mpd, 100, 1); % Minutes after midnight
y = randi(50, 100, 1);

Create a timetable.

tt = timetable(T + minutes(x), y);

Define a function to count number of y values greater than 20 in each bin.

fh = @(data) nnz(data > 20);

Retime the original timetable using the method fh, binning the data by hour.

tt2 = retime(tt, 'hourly', fh);

Let's see how many hours had at least 1 value in the original data greater than 20. Just show the first couple.

H = head(tt2)

Look at the values in tt that fell into the first bin in tt2.

tt(timerange(H.Time(1), H.Time(2)), :)

Since I set rng default at the top, you should see what I do: three values that exceed 20 and one that doesn't in the time between midnight and 1 AM. So the count of 3 in tt2 for that bin is correct.

ans =
  4×1 timetable
            Time            y 
    ____________________    __
    03-Sep-2019 00:52:00    49
    03-Sep-2019 00:46:00    46
    03-Sep-2019 00:50:00     8
    03-Sep-2019 00:18:00    41

Bin 6 (5 AM to 6 AM) of tt has no values of y greater than 20, so the corresponding element in tt2 is 0.

>> tt(timerange(H.Time(6), H.Time(7)), :)
ans =
  2×1 timetable
            Time            y
    ____________________    _
    03-Sep-2019 05:23:00    6
    03-Sep-2019 05:51:00    1
>> tt2(6, :)
ans =
  timetable
            Time            y
    ____________________    _
    03-Sep-2019 05:00:00    0

Best Answer

Related Solutions

MATLAB: Sequence of date with no leap year

MATLAB: I have a time table with a year’s worth of current measurements. (Time format is Year-Month​-Day-Hour-​Min-Sec) How to take each day and count how many times the current was over a specific reading that day.

Related Question

MATLAB: I have a time table with a year’s worth of current measurements. (Time format is Year-Month-Day-Hour-Min-Sec) How to take each day and count how many times the current was over a specific reading that day.