Prophet is an open-source forecasting tool released by Facebook’s Core Data Science team in 2017, described by Sean J. Taylor and Benjamin Letham in their paper “Forecasting at scale” (The American Statistician, 2018). The project page and library are maintained at facebook.github.io/prophet, with implementations in both R and Python.
Prophet fits a forecast as the sum of a few interpretable pieces: a trend that can bend at automatically detected changepoints, repeating seasonal patterns at yearly, weekly, and daily cycles, and the effect of holidays and special events that analysts can list. This additive structure means a non-expert can get a reasonable forecast on typical business time series, such as daily web traffic or sales, and can adjust it using intuitive knobs rather than statistical jargon. The design goal stated in the name was to let a large organization produce many forecasts at scale without a forecasting specialist tuning each one.
Prophet is not always the most accurate method on hard problems, and dedicated forecasters often beat it, but its accessibility made it one of the most popular forecasting libraries in industry.
Why business readers should care: Prophet lets ordinary analysts produce defensible forecasts for planning and budgeting without needing a time-series PhD on staff.