Replace Nan with previous numbers

9 views (last 30 days)
Hello there , I have a Variable of type double with natural gas Prices that contains Nans for dates that I don't have data and i want to replace Nans with the previous day known price. I tried function "fillts" but i didn't got what I wanted to . I appreciate any help Thanks in advance
data = [...
7.51
NaN
NaN
7.61
NaN
NaN
7.85];

Accepted Answer

Andrei Bobrov
Andrei Bobrov on 5 Jul 2012
Edited: Andrei Bobrov on 5 Jul 2012
t = ~isnan(data);
a = data(t);
out = a(cumsum(t));
ADD on Aggelos's comment
t = ~isnan(data);
idx = cumsum(t);
idx(~idx) = 1;
a = data(t);
out = a(idx);
  2 Comments
aggelos
aggelos on 5 Jul 2012
Thanks very much Andrei ,but i get the following error...
>> t = ~isnan(NEdata.NaturalGas); a = NEdata.NaturalGas(t); out = a(cumsum(t)); Subscript indices must either be real positive integers or logicals.
aggelos
aggelos on 7 Jul 2012
Thank you very much Andrei ... it works just fine . Thanks again ...keep on the great job

Sign in to comment.

More Answers (3)

Image Analyst
Image Analyst on 5 Jul 2012
Try this:
m = [nan 2 3 4 nan 6 7 nan 9] % Sample data
indexes = find(isnan(m))
if ~isempty(indexes)
% Handle case where first index is nan - get rid of it.
if indexes(1) == 1
m = m(2:end);
indexes = indexes(2:end) - 1
end
% Take prior index if element is nan.
m(indexes) = m(indexes-1)
end
  2 Comments
aggelos
aggelos on 5 Jul 2012
I get the same error as with Andrei's answer :Subscript indices must either be real positive integers or logicals. Any suggestions... Thank you very much
Image Analyst
Image Analyst on 5 Jul 2012
Edited: Image Analyst on 5 Jul 2012
You did something wrong, because I just copied and pasted my code on another computer and it worked great here also.
Plus you still need to say what you want done in case the first element is a nan. andrei and my vectorized solutions chose to handle it in different ways, while John non-vectorized code assumed that you would never have such a case (a non-robust, risky assumption). andrei's code seems to be even more robust than mine because his handles sequences of multiple nans in a row while mine doesn't. But his takes the second element for the first output element if the first input element is a nan, while mine just throws it away and shortens the array - that's why I asked what you want to do in such a case.

Sign in to comment.


John
John on 5 Jul 2012
try the following:
x = isnan(data);
for i = length(data)
if x(i) == 1
data(i) = data(i-1);
end
end

Chad Greene
Chad Greene on 1 Nov 2014
repnan can do it by
data = repnan(data,'previous');

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!