Resample/Match random interval data to interval timeseries

Question

loes visser 2017-3-9

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/328977-resample-match-random-interval-data-to-interval-timeseries

回答： Peter Perkins 2017-3-9

Hi!

I have multiple arrays with data that that is collected randomly in time. Like;

a = 
01-10-2016 00:00:00,  5
01-10-2016 00:00:03,  10
01-10-2016 00:00:08,  4
01-10-2016 00:00:10,  90
01-10-2016 00:00:11,  2

I would like to insert empty cells (NaN) between the measurements. Or match the data to a new timeserie, so I would get;

b = 
  01-10-2016 00:00:00,  5
  01-10-2016 00:00:01,  NaN
  01-10-2016 00:00:02,  NaN
  01-10-2016 00:00:03,  10
  01-10-2016 00:00:04,  NaN
  01-10-2016 00:00:05,  NaN
  01-10-2016 00:00:06,  NaN
  01-10-2016 00:00:07,  NaN
  01-10-2016 00:00:08,  4
  01-10-2016 00:00:09,  NaN
  01-10-2016 00:00:10,  90
  01-10-2016 00:00:11,  2

How can I accomplish this? interp or interp1 will fill every timestamp with a value, and this is not the goal.

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 2017-3-9

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/328977-resample-match-random-interval-data-to-interval-timeseries#answer_258034

在 MATLAB Online 中打开

If you have access to R2016b, use a timetable:

>> Time = {'01-10-2016 00:00:00'; '01-10-2016 00:00:03'; '01-10-2016 00:00:08'; '01-10-2016 00:00:10'; '01-10-2016 00:00:11'};
>> Time = datetime(Time,'InputFormat','MM-dd-yyyy HH:mm:ss');
>> x = [5; 10; 4; 90; 2];
>> tt = timetable(Time,x)
tt =
  5×1 timetable
            Time            x 
    ____________________    __
    10-Jan-2016 00:00:00     5
    10-Jan-2016 00:00:03    10
    10-Jan-2016 00:00:08     4
    10-Jan-2016 00:00:10    90
    10-Jan-2016 00:00:11     2
>> tt = retime(tt,'secondly')
tt =
  12×1 timetable
            Time             x 
    ____________________    ___
    10-Jan-2016 00:00:00      5
    10-Jan-2016 00:00:01    NaN
    10-Jan-2016 00:00:02    NaN
    10-Jan-2016 00:00:03     10
    10-Jan-2016 00:00:04    NaN
    10-Jan-2016 00:00:05    NaN
    10-Jan-2016 00:00:06    NaN
    10-Jan-2016 00:00:07    NaN
    10-Jan-2016 00:00:08      4
    10-Jan-2016 00:00:09    NaN
    10-Jan-2016 00:00:10     90
    10-Jan-2016 00:00:11      2

Prior to 16b, it should be pretty easy to do that using a datetime in a table, and probably something like ismember, or perhaps interp1 on Time and x if you want to fill in those NaNs. Hope this helps.