NeuralProphet Tutorial (I) — Trend + Seasonality + Holidays

Chris Kuo/Dr. Dataman
17 min readJan 9, 2024

Many of you may have heard about the open-source Prophet module for time series forecasting. Prophet sets a paradigm in time series forecasting community. Its interpretability of the forecasts as well as interactive user-interface have been welcomed by many professionals. I covered Prophet In the series “Business Forecasting with Facebook’s Prophet”. Now, with the release of the NeuralProphet module, I am even more excited to write the tutorial series to introduce NeuralProphet to you. If you have never used Prophet before, you can click “Business Forecasting with Facebook’s Prophet”.

You may wonder what’s new in NeuralProphet, or even, what’s new in Prophet. In the tutorial series, I will start with a re-cap of Prophet and introduce NeuralProphet. Then I will add modules in a sequence. Neural Prophet inherits the trend, seasonality, holidays & events of Prophet, and expands to the auto-regressive, lagged regressors, and future regressors components. Each component can be experimented and optimized as I will show you in my tutorials.

I organize the tutorial series into digestible sizes like the following:

--

--

Chris Kuo/Dr. Dataman

The Dataman articles are my reflections on data science and teaching notes at Columbia University https://sps.columbia.edu/faculty/chris-kuo