Prophet

Warnings

• breaking The official package name changed from `fbprophet` to `prophet`. Projects still using `fbprophet` will encounter import errors and miss out on updates.
  Fix: Update your import statements from `from fbprophet import Prophet` to `from prophet import Prophet`. If you were using `pip install fbprophet`, switch to `pip install prophet`.
• gotcha Installation of `prophet` often requires a C++ compiler and specific environment setup due to its `cmdstanpy` dependency. Users on Windows or macOS might need to install development tools (e.g., Build Tools for Visual Studio, Xcode Command Line Tools).
  Fix: Ensure a compatible C++ compiler is installed and accessible in your PATH. Refer to the `cmdstanpy` documentation for detailed installation instructions for your operating system (e.g., `https://mc-stan.org/cmdstanpy/installation.html`).
• gotcha Prophet expects the 'ds' column to contain naive (timezone-unaware) datetime objects, ideally in UTC. Providing timezone-aware datetimes can lead to unexpected behavior or errors.
  Fix: Before passing your DataFrame to Prophet, ensure the 'ds' column is converted to timezone-naive datetimes. You can use `df['ds'] = df['ds'].dt.tz_localize(None)` to remove timezone information.
• gotcha Prophet has strict dependencies on specific versions of `pandas` and `numpy`. Version conflicts can cause runtime errors or unexpected behavior. While recent versions (1.3.0) aim to broaden compatibility, it remains a common issue.
  Fix: Always install `prophet` in a clean virtual environment. If issues arise, try to `pip install "pandas<2"` or check the `prophet` documentation/changelog for explicitly supported `pandas` and `numpy` versions.

Install

• pip install prophet Standard installation

Imports

• Prophet

  from prophet import Prophet

  The library was renamed from 'fbprophet' to 'prophet'. Using 'fbprophet' will result in an ImportError as it is deprecated and no longer maintained.

Quickstart

This quickstart demonstrates how to create a simple Prophet model, fit it to a Pandas DataFrame, generate a future DataFrame, and make predictions. The input DataFrame must have columns named 'ds' (datetime) and 'y' (numerical value).

import pandas as pd
from prophet import Prophet

# Create a sample DataFrame with 'ds' (datetime) and 'y' (value) columns
data = {
    'ds': pd.to_datetime(['2017-01-01', '2017-01-02', '2017-01-03', '2017-01-04', '2017-01-05',
                          '2017-01-06', '2017-01-07', '2017-01-08', '2017-01-09', '2017-01-10']),
    'y': [10, 12, 15, 13, 17, 18, 20, 22, 25, 23]
}
df = pd.DataFrame(data)

# Instantiate and fit the Prophet model
model = Prophet()
model.fit(df)

# Create a future DataFrame for predictions (e.g., next 3 days)
future = model.make_future_dataframe(periods=3)

# Make predictions
forecast = model.predict(future)

# Print the last few rows of the forecast
print(forecast[['ds', 'yhat', 'yhat_lower', 'yhat_upper']].tail())