I have been facing a wall after doing a forecast of wind speed time series data using ARIMA with python.
I have result with a nrmse growth going from 2% to 15% and now what I want is to use kalman filter to reduce my forecast errors and have better result.
1/Actually I don't know even after lot of readings what is doing what. My main goal is to improve my forecast results (that I obtained using ARIMA). Can you enlighten me on Kalman Filter for forecasting univariate time series?
2/ I also want to know how I can know if a time series is non linear or not ? The definition on various papers does not allow me to write a code based on them. Thanks for your considerations?