Can interp1 function be used when the time interval of data isn't uniform?

In the interp1 function, yi = interp1(x,Y,xi,method), I have x as time vector with non-uniform interval. But the length of x and Y vectors are same. Can I use the interp1 function to interpolate the data?

Answer 1

[José-Luis]

Yes, but the longer your interval the more uncertain your results.

[Swasti Khuntia]

Thanks. My time vector size is 2482175x1. And, when I compare the results between linear and spline methods, it's totally different.

[José-Luis]

There are a few different methods for interpolation, that will not produce the same results. For more info on them:

doc interp1

Cheers!

[Walter Roberson]

All line segments have the same number of mathematical points ("the infinity of real numbers"), so the potential mathematical variability over any line segment is the same as for any other line segments. One cannot generalize about uncertainty over different intervals without knowing the process responsible for the variability.

As a quick example, the uncertainty of the sinc() function over short intervals near 0 is much greater than the uncertainty over long intervals at large x. Larger intervals are not always more uncertain.

[José-Luis]

I see your point. It depends on what you are modelling or measuring of course. However, even if you (think) you know your process, there is the problem of unknown unknowns. Unless the system is a very simple one, things will arise that will surprise you, therefore the more uncertainty the larger the interval. And if you are sure of how it works, then why would you need to measure? It all depends on your perspective (as always), and it is an interesting discussion, but maybe overkill for the question of the op.

Answer 2

[Matt Fig]

When I have a question like this, I just do a little experiment.

x = sort(rand(1,30)*2*pi); % Non-uniform data
y = sin(x);  % Underlying function
xi = 0:.25:2*pi;  % Interpolation points
yi = interp1(x,y,xi,'spline');
plot(x,y,'*r',xi,yi,'b')