All Tools
N
DataFreeOpen Source
NEURALPROPHET
Interpretable neural time series forecasting inspired by Prophet
MIT
ABOUT
Classical Prophet is easy to explain but weak on high-frequency series, while deep forecasters are accurate and opaque. NeuralProphet keeps trend, seasonality, and holiday components you can plot, then adds PyTorch AR-Net terms so teams iterate on interpretable forecasts without jumping to a black-box neural stack.
INSTALL
pip install neuralprophetINTEGRATION GUIDE
1. Forecast sub-daily metrics with weekly and yearly seasonality you can inspect
2. Add holidays and changepoints while keeping a Prophet-like decomposition
3. Iterate human-in-the-loop models with plots of trend, seasonality, and residuals
4. Train autoregressive neural components on long high-frequency series
TAGS
pythonpytorchtime-seriesforecastingprophetinterpretableopen-source