Exclude the NaN, 0, empty and Inf values ​​from the analysis.

When calculating the PE value, I would like it not to calculate when Ia_future = 0, Nan or Inf.

I believe the way I did it, it's still calculating. Because some PE values ​​are still Inf.

Or if there was some way to exclude those values. The problem is that I need to plot a (t,PE) graphic and if I exclude some PE values ​​the two will have different dimensions and I will not be able to analyze the graph..

for n = 4:size(t,1)
    X = [Ia(n-1,1) Ia(n-2,1) ; Ia(n-2,1) Ia(n-3,1)];
    future = [Ia(n,1) ; Ia(n-1,1)];
    C = X\future;
    Ia_future(n,1) = C(1,1)*Ia(n,1)+C(2,1)*Ia(n-1,1);
    if (isnan(Ia_future(n, 1)) || isinf(Ia_future(n,1) || isempty(Ia_future(n,1) || Ia_future(n,1)==0))) %|| %(isnan(p(n, 1)) || p(n, 1) == 0)
            continue
    end
    PE(n,1)=(Ia(n,1)+Ia_future(n,1))/(2000/5);
end

Answer 1

[Benjamin Thompson]

Vectorize your calculations using index vectors. For example:

>> Test = [0 1 inf NaN]

Test =

0 1 Inf NaN

>> Inan = isnan(Test)

Inan =

1×4 logical array

0 0 0 1

>> Iinf = isinf(Test)

Iinf =

1×4 logical array

0 0 1 0

>> Igood = ~isinf(Test) & ~isnan(Test)

Igood =

1×4 logical array

1 1 0 0

Then you can calculate PE as a function of Ia_future outside the for loop, something like:

PE(Igood,1)=(Ia(Igood,1)+Ia_future(Igood,1))/(2000/5);

Only the rows of PE corresponding to where Igood is one will be updated. You may need to calculate the index vector looking at both Ia and Ia_future if they can have bad values in different spots.

[Luccas S.]

Cool, I tested it here and it worked, you can do several things with this logic, I think.

If I wanted to ignore the values of PE>1 and PE<0 would it be possible to do something like that too?

Thanks!

[Benjamin Thompson]

Yes you can define an index vector using any kind of comparison test.

Answer 2

[Image Analyst]

Is this helpful:

data = [0, 9, inf, NaN, 42];
mask = (data ~= 0) & isfinite(data)

mask = 1×5 logical array
   0   1   0   0   1

extractedData = data(mask)

extractedData = 1×2

9 42

Using isfinite() takes the place/combines both ~isinf() and ~isnan() all into one simple function.

[Luccas S.]

Okay, I didn't know about this isfinite function. Thanks too !