Closed RobinloveCode closed 5 years ago
I write a multivariate high order model by myself.
There is only one multivariate high order method on pyFTS, the cmvfts.ClusteredFTS (https://pyfts.github.io/pyFTS/build/html/pyFTS.models.multivariate.html#module-pyFTS.models.multivariate.cmvfts) and a subclass granular.GranularWMVFTS which is easier to use.
There is a small demonstration of this method on the following link:
A simple guide to use the above methods:
An example code can be found here: https://colab.research.google.com/drive/1bTaMcxLP0I_maqJdZukaYeaP8sqfoZA2
For this model there is a new forecasting type, the 'multivariate'. When you inform type='mulivariate' in the predict method the return will be a pandas dataframe with the forecasts of all variables, not only the target variable. This makes this method a MIMO (multiple input multiple output) instead of the usual MISO (multiple input single output) model of the others multivariate methods on pyFTS.
WARNING! If in the variables there is a deterministic one (for example date parts, etc) use generators!
The above method is under publishing, as soon as the paper become available I will share it here.
If you want to, you can include your model on pyFTS library!
Hi Dr.
Is there any tutorial about high-order multi-variate Fuzzy time series model?
Thanks.