Open slavakurilyak opened 6 years ago
@bukosabino We can build a statistical model to predict the trendline for daily and non-daily data using prophet (see above). Then, we can create an ensemble model (combining the prophet model with a machine learning model) or use the prophet model as a feature for the machine learning model.
I have a lot of problems to integrate fbprophet on the system: https://github.com/enigmampc/catalyst/issues/388
I'm not alone: https://github.com/enigmampc/catalyst/issues/269
They need update pandas version as soon as possible!
Goal
As a developer, I want to implement nonlinear regression models, so that I can develop statistical indicators.
Consider
Consider using facebook's prophet library, which uses nonlinear regression models with "interpretable parameters that can be intuitively adjusted by analysts with domain knowledge about the time series." (Prophet's Paper, 2017).
Consider leveraging work of others, whom performed Bitcoin analysis using prophet: Bishoph, doubledare704, Joshuaek