> ML_ALGORITHM // GENERALIZED-ADDITIVE-PROPHET_v1.0

Prophet (Additive Decomposable Time Series)

Robust business forecasting tool developed by Meta that frames time series modeling as a curve-fitting problem combining trend, Fourier seasonality, and holiday terms.

Decomposable Generalized Additive Modelstime-series-forecastinghigh-intrinsicsmall (<1k)

Back to All Algorithms

Computational Complexity

Training Complexity:O(changepoints * Fourier_order * n)

Inference Complexity:O(horizon)

Hardware Profile

CPU Friendly:Yes

Requires GPU:No

Memory Footprint:low

Interpretability & Data

Interpretability Tier:high-intrinsic

Training Data Needs:small (<1k)

Interpretability Assessment

Provides clear decomposed component plots for trend, weekly seasonality, annual seasonality, and specific holidays.

Suitable Tasks & Supported Modalities

Suitable Tasks:

business metric forecastingcapacity planning

Supported Modalities:

time-series

Implementing Libraries

ProphetMeta Open Source · v1.1.5

View Spec

stan

Foundational Literature

Forecasting at Scale (Prophet)Sean J. Taylor, Benjamin Letham (2018) · The American Statistician

Common Pitfalls & Warnings

Prophet does not model autoregressive error lags, resulting in autocorrelated residuals and poor short-term 1-step-ahead forecasts