Using "isfinite" to skip NAN data

Hi all,

I have 2 set of data, each of which is 31*12 (31 for number of days, and 12 for months)

Set1 = T3_All (%x-values)
Set2 = Tmax_All (%y values)

Since we don't have more than 28 or 29 days for some months, the data is filled with NAN.

when i use ployfit (code below), I am getting NaN for all values. I think that is because my arrays include NaN values.

How can I get the polyfit function pass the data that contains nan, using isfinite?

Here is my code:

% Assigning arrays to data
T3_All=AllT3_Import;
Tmax_All=AllTmax_Import;
% Determining coefficients
coeff1 = polyfit(T3_All,Tmax_All,1); 
And the results I am getting:
coeff1 =
   NaN   NaN

P.S. I have posted this question before, although the answer provided was excellent, but it solved the problem using "mask". We are required to use "isfinite".

Many thanks

Answer 1

[Bruno Luong]

isvalid = isfinite(T3_All) & isfinite(Tmax_All);
coeff1 = polyfit(T3_All(isvalid),Tmax_All(isvalid),1)

You would recognize it is a mask, if your professor doesn't like it, ask why, because the mo mask solution is slower and less visible.

ikeep = find(isfinite(T3_All) & isfinite(Tmax_All));
coeff1 = polyfit(T3_All(ikeep),Tmax_All(ikeep),1)

Answer 2

[KSSV]

You have two options:

1. Use interpolation to fill the nan's and then use polyfit.

2. Remove NaN's using isnan and then fit the data.

Answer 3

[D.J]

Perfect results ! Thanks a lot Bruno !