search

facebook / prophet

Tool for producing high quality forecasts for time series data that has multiple seasonality with linear or non-linear growth.
https://facebook.github.io/prophet
MIT License
18.38k stars 4.52k forks source link

' .predict' method takes too much time to run #641

Closed ZED711 closed 6 years ago

ZED711 commented 6 years ago

Hello everyone. I'm currently doing a forecast with 3 months of data and trying to make a forecast of 1 month. However, my program takes too much time to run.

my_model = Prophet(interval_width=0.8,changepoint_prior_scale = 0.1) df_future = my_model.make_future_dataframe(periods=31, freq="D")

The program takes too much time on this line of code:

forecast = my_model.predict(df_future_final)

I've implemented Fourier regressors to model the weekends and removed outliers, not sure if that might be a problem in terms of optimization, but I don't think so. Btw, the data is weekly periodic, should I put weekly_seasonility = True?

ispmarin commented 6 years ago

How much is "too much time to run"?

ZED711 commented 6 years ago

I let it for hours and it didn't run

ispmarin commented 6 years ago

Can you provide a data sample and the code?

ZED711 commented 6 years ago

Nevermind, I have solved the problem.

markrazmandi commented 6 years ago

What was your issue? I have experienced the same with specific parameter combinations.