ourownstory/neural_prophet

Health v2 · maintenance only

Adoption not part of the score

4295 stars · 514 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

NeuralProphet is a Python library for interpretable time series forecasting built on PyTorch, combining neural networks with traditional time-series algorithms inspired by Facebook Prophet and AR-Net. It supports trends, seasonality, autoregression, regressors, events, and uncertainty quantification with a simple, human-in-the-loop modeling workflow.

Use cases

• forecast daily sales from historical time series data
• predict energy load or demand for a facility
• model seasonality and trends in sub-daily time series
• migrate an existing Facebook Prophet forecasting workflow to a neural approach
• build interpretable forecasting models with autoregression and lagged regressors
• quantify uncertainty in time series forecasts

When to choose

• you want an easy-to-learn, interpretable forecasting framework with a Prophet-like API
• your time series is high-frequency (sub-daily) and spans at least two full periods/years
• you want to iterate on model components like trend, seasonality, and events in a human-in-the-loop fashion
• you prefer a PyTorch-based model that integrates with the Python ML ecosystem

When to avoid

• you need the most accurate out-of-the-box forecasts without tuning or iteration
• your data is low-frequency with only one or fewer observed periods
• you need a fully mature, production-hardened tool - the project notes it is still in beta
• you require non-Python environments or non-PyTorch deep learning backends

Facets

library · maturity active

machine-learning deep-learning data-science machine-learning data-science time-series artificial-intelligence python forecasting time-series pytorch prophet autoregression seasonality interpretability

Member repositories

For agents

markdown · JSON · MCP: product_card(name="ourownstory/neural_prophet")