Forecasting time series with multiple seasonal patterns using NeuralProphet

NeuralProphet was inspired by the Prophet library and Autoregressive Neural Network (AR-Net) to bring a new implementation, leveraging deep neural networks to provide a more scalable solution.

Prophet was built on top of PyStan, a Bayesian inference library, and is one of the main dependencies when installing Prophet. Conversely, NeuralProphet is based on PyTorch and is as used as the deep learning framework. This allows NeuralProphet to scale to larger datasets and generally provides better accuracy than Prophet. Like Prophet's method, NeuralProphet performs hyperparameter tuning and fully automates many aspects of time series forecasting.

In this recipe, you will compare the results using NeuralProphet against Prophet.

Getting ready

It is recommended to create a new virtual Python environment this way, you can install all the required dependencies without any conflicts or issues with your...