Find date time value that corresponds to maximum daily value

I have a dataset from a tide gauge that measures tide elevation every 6 minutes for 3 and a half years which adds up to over 200,000 data values. I was able to extract the maximum daily values, but I also want to know the corresponding date time values. So I have a vector of date time values, a vector of tide elevation values of the same length, and a subset of maximum daily tide values, and I want the corresponding date time values to that maximum daily tide subset. Can anyone help me? I can provide my dataset if that helps.

Answer 1

[Peter Perkins]

I'm not sure retime will do all of what's asked for: it only works on one variable at a time. Getting the max tide for each day is easy, getting the specific time during each day when that happened is trickier. You can do it with retime, but I think it would take a few tricks, unless I'm not seeing something. Here's a version that uses a grouped rowfun. It requires one to convert the timetable (temporarily) to a table, because the row times of a timetable cannot curently be specified as an input to the function being applied.

>> X = rand(10,1);
>> tt = timetable(X,'RowTimes',datetime(2019,1,23,4:8:76,0,0))
tt =
  10×1 timetable
            Time               X   
    ____________________    _______
    23-Jan-2019 04:00:00    0.95974
    23-Jan-2019 12:00:00    0.34039
    23-Jan-2019 20:00:00    0.58527
    24-Jan-2019 04:00:00    0.22381
    24-Jan-2019 12:00:00    0.75127
    24-Jan-2019 20:00:00     0.2551
    25-Jan-2019 04:00:00    0.50596
    25-Jan-2019 12:00:00    0.69908
    25-Jan-2019 20:00:00     0.8909
    26-Jan-2019 04:00:00    0.95929
>> t = timetable2table(tt);
>> t.Day = dateshift(t.Time,'start','day');
t =
  10×3 table
            Time               X               Day         
    ____________________    _______    ____________________
    23-Jan-2019 04:00:00    0.95974    23-Jan-2019 00:00:00
    23-Jan-2019 12:00:00    0.34039    23-Jan-2019 00:00:00
    23-Jan-2019 20:00:00    0.58527    23-Jan-2019 00:00:00
    24-Jan-2019 04:00:00    0.22381    24-Jan-2019 00:00:00
    24-Jan-2019 12:00:00    0.75127    24-Jan-2019 00:00:00
    24-Jan-2019 20:00:00     0.2551    24-Jan-2019 00:00:00
    25-Jan-2019 04:00:00    0.50596    25-Jan-2019 00:00:00
    25-Jan-2019 12:00:00    0.69908    25-Jan-2019 00:00:00
    25-Jan-2019 20:00:00     0.8909    25-Jan-2019 00:00:00
    26-Jan-2019 04:00:00    0.95929    26-Jan-2019 00:00:00
>>> rowfun(@myFun,t,'GroupingVariable','Day','OutputVariableNames',{'MaxTime' 'MaxX'})
ans =
  4×4 table
            Day             GroupCount          MaxTime            MaxX  
    ____________________    __________    ____________________    _______
    23-Jan-2019 00:00:00        3         23-Jan-2019 04:00:00    0.95974
    24-Jan-2019 00:00:00        3         24-Jan-2019 12:00:00    0.75127
    25-Jan-2019 00:00:00        3         25-Jan-2019 20:00:00     0.8909
    26-Jan-2019 00:00:00        1         26-Jan-2019 04:00:00    0.95929

And then convert that back to a timetable. I think you can also do the above with splitapply.

myFun looks like this:

function [tmax,xmax] = myFun(t,x)
[xmax,imax] = max(x,[],1);
tmax = t(imax);

[Peter Perkins]

My original post left out two key parts of the code, t.Day = ..., and the definition of myFun. I've added them.

[SaaraL]

I am trying to do the same thing, but on many stations at once. Is that possible? Tried to do it on one station, but it gives me an error saying that there are too many input arguments.

[Peter Perkins]

Add a grouping variable for Station to what I have shown. "By station and day" is no different than "by day", just one more grouping variable.

Answer 2

[Star Strider]

It will likely be best to create a timetable if your data, then use the retime function to calculate the maximum daily value. You might find the documentation section on Synchronize and Aggregate Data Values helpful to determine the time the maximum occurred.