---
id: prophet
version: "1.3.0"
license: MIT
license_treatment: permissive
maintenance: active
---

# prophet — Automatic Forecasting Procedure

License: permissive · Maintenance: active · Downloads: 7.0M/mo

## What it is and what it does

Prophet is a time-series forecasting library built by Facebook's Core Data Science team that automates the process of fitting trends and seasonal patterns to historical data. It decomposes a time series into trend, seasonality (yearly, weekly, daily), and holiday components, then uses an additive model to project future values. The library is designed to work well with data that has strong seasonal effects and multiple seasons of history, and it tolerates missing data and trend shifts gracefully.

You provide Prophet with a pandas DataFrame containing a time column ('ds') and a target column ('y'), call fit() to train the model, then use make_future_dataframe() and predict() to generate forecasts. The package depends on cmdstanpy (which wraps Stan, a probabilistic programming language), numpy, pandas, matplotlib, and a few utility libraries. Installation requires a C++ compiler because Stan code is compiled during setup, which adds medium friction but is a one-time cost.

Use it for:

- Forecast sales, revenue, or demand over the next quarter or year for business planning and inventory management.
- Predict website traffic, user engagement, or API request volumes to capacity-plan infrastructure.
- Estimate energy consumption or utility demand by hour, day, or season for grid management.
- Project staffing needs or call-center volume based on historical patterns with holiday adjustments.
- Monitor and forecast metrics in monitoring systems where you need to detect anomalies relative to expected seasonal trends.

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

Prophet forecasts time series data using an additive model that combines non-linear trends with seasonal patterns (yearly, weekly, daily) and holiday effects, handling missing data and outliers robustly.

Yes, if you need automatic time-series forecasting with built-in seasonal and holiday handling. Prophet is well-maintained, actively used in production (7M+ monthly downloads), has no known vulnerabilities, and the MIT license poses no restrictions. The medium install friction (C++ compilation via cmdstanpy) is a one-time cost. Install it if your data has clear seasonal patterns and you want a library that requires minimal tuning; skip it if you need ultra-low latency inference or forecasting on non-seasonal data.

## Install

pip install prophet
uv add prophet
poetry add prophet

## Installing prophet

Before you install: Medium install friction due to compiled dependencies (cmdstanpy). The package is actively maintained with recent releases, supports Python 3.7–3.12, and provides pre-built wheels for macOS (Intel and ARM), Linux, and Windows, reducing build-from-source burden on most platforms.

License in practice: MIT license is permissive; you can use, modify, and distribute Prophet freely with minimal restrictions, making it suitable for both open-source and commercial projects.

Quickstart:

pip install prophet
from prophet import Prophet
m = Prophet()
m.fit(df)  # df is a pandas.DataFrame with 'y' and 'ds' columns
future = m.make_future_dataframe(periods=365)
m.predict(future)

cmdstanpy requires a C++ compiler toolchain; on Windows, Visual Studio Build Tools or MinGW may be needed. Installation can take several minutes due to Stan compilation.

Verify before relying:

- Whether the package's forecasting accuracy is competitive with or superior to alternatives like statsmodels or other modern time-series libraries.
- Real-world performance on datasets with extreme seasonality or non-standard patterns beyond the design assumptions.
- Whether the holiday effects module covers all regions and holidays relevant to your use case.

## Package facts

- License: MIT (permissive)
- Python support: supports_current
- Install friction: medium
- Maintenance: active
- Downloads: 7.0M/month (top 5,000 on PyPI)
- Known vulnerabilities: none known

## Tags

time series forecasting, seasonal trend prediction, automatic forecasting, holiday effects forecasting, time series decomposition, multi-seasonal forecasting, robust time series prediction, forecasting, time-series, seasonality

[View on SkillFed](https://skillfed.io/packages/prophet) · [View on PyPI](https://pypi.org/project/prophet/)