Insert new rows of NaN for missing timestamps in time series.

Question

Eric Escoto 2019-1-31

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/442551-insert-new-rows-of-nan-for-missing-timestamps-in-time-series

回答： Peter Perkins 2019-2-7

Hello, I have a time series that is missing some data. In the places where there is a data gap I would like to insert rows of NaN values at the same incremetal time step. The data is organized by a datenum.

Here is the current script I'm using to get the timeseries.

Thanks!

% Read the data.

[num, txt, raw] = xlsread('SFL_2018.xlsx'); % Creates three variables. A number (double), raw (cell), and txt (cell).

% Note that the 'txt' and 'raw' variables contain the decriptions of the columns.

% Rename the numerical matrix and remove the upper 4 rows that originally contained the data headers.

SFL2018 = num(5:end, :);

% Change the first column (excel datenum value) to MATLAB datenum value.

datevec = datetime(SFL2018(:,1),'ConvertFrom','excel');

date_num = datenum(datevec);

SFL2018 = [date_num SFL2018(:, 2:end)];

% Rename the 'txt' cell to 'header' and clear unused variables.

header = txt(1:4,:);

clearvars('num', 'txt', 'raw', 'date_num');

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

Peter Perkins 2019-2-7

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/442551-insert-new-rows-of-nan-for-missing-timestamps-in-time-series#answer_359964

在 MATLAB Online 中打开

retime "adjusts" your data to a new set of row times. In general, "adjust" means things like interpolate, or aggregate, or take the neaest value, but by default it fiills with missing values. So

>> tt = timetable([1;2;5;6],'RowTimes',datetime(2019,2,[1;2;5;6]))
tt =
  4×1 timetable
       Time        Var1
    ___________    ____
    01-Feb-2019     1  
    02-Feb-2019     2  
    05-Feb-2019     5  
    06-Feb-2019     6  
>> retime(tt,datetime(2019,2,1:6))
ans =
  6×1 timetable
       Time        Var1
    ___________    ____
    01-Feb-2019      1 
    02-Feb-2019      2 
    03-Feb-2019    NaN 
    04-Feb-2019    NaN 
    05-Feb-2019      5 
    06-Feb-2019      6 

Or in this simple case

>> retime(tt,'daily')
ans =
  6×1 timetable
       Time        Var1
    ___________    ____
    01-Feb-2019      1 
    02-Feb-2019      2 
    03-Feb-2019    NaN 
    04-Feb-2019    NaN 
    05-Feb-2019      5 
    06-Feb-2019      6 

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。

Answer 2

Peter Perkins 2019-1-31

1
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/442551-insert-new-rows-of-nan-for-missing-timestamps-in-time-series#answer_358952

I recommend you use a timetable. Read your data in using readtable, then use table2timetable. The retime function makes what you want one line.

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Eric Escoto 2019-2-1

Thanks, Peter

This method doesn't seem to create new rows and insert them into the SFL2018 array, creating a new array. If it does, I don't quite get how that would happen. Could you provide an example?

-Eric

请先登录，再进行评论。