loop subsampling a timetable for each year for annual max values

Question

NOtway 2023-5-27

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1974434-loop-subsampling-a-timetable-for-each-year-for-annual-max-values

评论： Peter Perkins 2023-6-5

I have 9 different daily-interval timetable datasets, each spanning ~100 years of data. I need to extract the maximum annual 5-day sum of variable values for each year, in each dataset.

I've figured out how I can get the maximum 5-day sum across the entire 100 year dataset:

v = Timetable_1of9.Var1 ; 
n = 5 ; 
N = length(v) ; 
sumn = zeros(1, N - n + 1);  % Pre-allocation
for i = 1:N - n + 1
  sumn(i) = sum(v(i:(i+n-1))) ;
end
[val,idx] = max(sumn) 

but I'm not sure of the most efficient way to do this for each year and each dataset. (I also need those values put back into 9 new timetables with each annual maximum assigned to 1-1-YYYY, to match in with later analysis I will be doing).

If there a way I can set up a batch or a another for loop to run through each? even if it just does 1 dataset at a time, I can easily copy the code 8 more times and change the input file. Doesn't need to be pretty, just needs to work.

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

NOtway 2023-5-31

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1974434-loop-subsampling-a-timetable-for-each-year-for-annual-max-values#answer_1247999

在 MATLAB Online 中打开

If anyone in the future is looking for an anser: I wrked it out myself. Its definitely not the most elegant solution, but it works so good enough for me

%% sum each 5 days from 1-Jan-2015
Aggregate_5day_1 = retime(Aggregate_1day, 'regular', 'sum', 'TimeStep', caldays(5));
% find annual max of these values
Aggregate_5day_1_max = retime(Aggregate_5day_1,"yearly", "max");
%% sum each 5 days from 2-Jan-2015
% delete 1-Jan-2015 from Aggregate_1day timetable
Aggregate_1day(1,:) = [];
Aggregate_5day_2 = retime(Aggregate_1day, 'regular', 'sum', 'TimeStep', caldays(5));
% find annual max of these values
Aggregate_5day_2_max = retime(Aggregate_5day_2,"yearly", "max");
%% sum each 5 days from 3-Jan-2015
% delete 2-Jan-2015 from Aggregate_1day timetable
Aggregate_1day(1,:) = [];
Aggregate_5day_3 = retime(Aggregate_1day, 'regular', 'sum', 'TimeStep', caldays(5));
% find annual max of these values
Aggregate_5day_3_max = retime(Aggregate_5day_3,"yearly", "max");
%% sum each 5 days from 4-Jan-2015
% delete 3-Jan-2015 from Aggregate_1day timetable
Aggregate_1day(1,:) = [];
Aggregate_5day_4 = retime(Aggregate_1day, 'regular', 'sum', 'TimeStep', caldays(5));
% find annual max of these values
Aggregate_5day_4_max = retime(Aggregate_5day_4,"yearly", "max");
%% sum each 5 days from 5-Jan-2015
% delete 4-Jan-2015 from Aggregate_1day timetable
Aggregate_1day(1,:) = [];
Aggregate_5day_5 = retime(Aggregate_1day, 'regular', 'sum', 'TimeStep', caldays(5));
% find annual max of these values
Aggregate_5day_5_max = retime(Aggregate_5day_5,"yearly", "max");
%% Find annual total max from the max of those 5 values for each year
Aggregate_5day_temp = synchronize(Aggregate_5day_1_max, Aggregate_5day_2_max, Aggregate_5day_3_max, Aggregate_5day_4_max, Aggregate_5day_5_max);
Wet3_7_5day = timetable(Aggregate_5day_temp.Time,Aggregate_5day_temp.Wet3_7_Aggregate_5day_1_max, Aggregate_5day_temp.Wet3_7_Aggregate_5day_2_max, ...
    Aggregate_5day_temp.Wet3_7_Aggregate_5day_3_max, Aggregate_5day_temp.Wet3_7_Aggregate_5day_4_max, Aggregate_5day_temp.Wet3_7_Aggregate_5day_5_max);
Wet3_7_5day_max_temp = max([Wet3_7_5day.Var1, Wet3_7_5day.Var2, Wet3_7_5day.Var3, Wet3_7_5day.Var4, Wet3_7_5day.Var5], [], 2);
Wet3_7_5day_max = array2timetable(Wet3_7_5day_max_temp, "RowTimes", Wet3_7_5day.Time);

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Peter Perkins 2023-6-5

Yes, this would be simple if it was "non-overlapping 5-day windows".

For what it's worth I think you could add 1-day,...4-day lagged versions of your data to the timetable and do the "five day max" part for the different lags all at once, and then collapse with a max across the different lags. Less code, anyway.

请先登录，再进行评论。