All data can be 'fed in' in parts. The filter will update itself and its estimation.
A model can be created and simulated. The results are plotted but the estimations can also be extracted if needed.
Filters a process using a Kalman filter.
Filters a process using a Particle filter.
An example of a simulation using non-linear model:
x := x/(1+x^2) + v y := x + w
w white noise is generated with code like below:
model = FilterModel(NormalDistribution(10, 1), NoiseDistribution(0.5), NoiseDistribution(0.5), lambda x: x/(1+x*x), lambda x: (1-x*x)/(x*x+1)**2, lambda x: x, lambda _: 1) simulate(n, model, 100, T)
The results would look like the following:
Look for more examples in the main code.