How to: Retime a timetable without omitting NaNs

Hi,

I'm trying to use the retime function on a 488 x 12 x 4 timetable, and calculate monthly sums from 488 daily values. The problem is, there are lots of missing values (NaNs) in my dataset, and since retime's 'sum' method omits NaNs, in months with lots of missing data, I receive falsly low sums. I tried using @sum, but I'm getting an error:

Unable to concatenate incompatible values returned by the function 'sum' when applied to groups in the variable 'Var1':

Dimensions of arrays being concatenated are not consistent.

Any help will be appreciated!

[Peter Perkins]

There's not currently any way to have "a 488 x 12 x 4 timetable", so I think you need to give a short example of what you have and the retime command you used that led to that error. It mazy be as simple as using @(x)sum(x,1).

[KAE]

I am getting this error too, perhaps due to irregularly sampled data so I don't have data in every bin. This is the only mention of the error I've found on the web. If you have figured it out, please post.

[Peter Perkins]

KAE, if you are using @sum, the problem is that the sum function returns an empty for empty input, and that can't be concatenated with the scalar value coming from the non-empty bins. That's why retime has it's own local version of sum, which you get by passing 'sum', not @sum. If you are doing something else, you need to give a short example of your data and what command you are using.