Timetable linear interpolation within a range

Question

Emanuel Valdes 2019-5-29

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/464653-timetable-linear-interpolation-within-a-range

评论： Adam Danz 2019-5-30

Data.mat

I have a timetable with measurments every 16 seconds. I need to linearly interpolate the missing values, only if there are less than 10 consecutive NaN rows, otherwise it have to remain as missing values.

This is the line for interpolate my data:

TT2 = retime(TT2,'regular','linear','TimeStep',seconds(16));

But that line interpolates everything. I suppose it should be approached with some kind of for loop, and an if statement making the interpolation only if the amount of consecutive NaN's is smaller than 10

I would really appreciate your help! thanks in advance!

3 个评论
显示 1更早的评论隐藏 1更早的评论

Emanuel Valdes 2019-5-29

row number 5 for example:

Adam Danz 2019-5-29

编辑：Adam Danz 2019-5-29

在 MATLAB Online 中打开

I see. I was expecting to see missing timestamps but clearly there are none (just missing data from some time stamps). I can update with some suggestions in a bit.

unique(diff(TT2.Fecha))
ans = 
  duration
   00:00:16
   
   % no missing time stamps

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Adam Danz 2019-5-29

1
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/464653-timetable-linear-interpolation-within-a-range#answer_377129

编辑：Adam Danz 2019-5-30

在 MATLAB Online 中打开

This solution has 3 steps:

determine which rows are within 10 or more consecutive rows of missing data
interpolate all missing data
replace the rows identified in step 1 with NaN values

% Find consecutive rows of NaN that exceed threshold number allowed
hasNan = all(isnan(TT2.data),2);   
dIdx = find(diff([0;hasNan;0]==1));     %rows that change 1/0
s1 = dIdx(1:2:end-1);                   %start indices of 1s
s2 = dIdx(2:2:end);                     %stop indices of 1s
keepNan = (s2-s1)>=10;                  %which segments have too many consecutive nans
hasNan(cell2mat(arrayfun(@(x1,x2)x1:x2, s1(~keepNan),s2(~keepNan)-1,'UniformOutput',false)')) = false; 
% Interp all missing values
TT2intrp = retime(TT2,'regular','linear','TimeStep',seconds(16));
% TT2intrp = fillmissing(TT2,'linear');  % This gives you the same results as your line above
% Replace the NaN values for >10 consecutive rows of missing data
TT2intrp.data(hasNan,:) = NaN; 