eurostat / prophet

Application of Facebook Prophet model (Python) for forecasting Eurostat monthly indicators

prophet

Applying Facebook Prophet model for forecasting Eurostat monthly indicators

This is a blind/dummy (no assumption whatsoever) application of Prophet automatic procedure for forecast estimates of Eurostat tour_occ_nim time-series on the number of "nights spent at tourist accommodation establishments" per month.

Description

(from the source webpage)

At its core, the Prophet procedure is an additive regression model with four main components (using Stan Bayesian approach, see reference below):

• a piecewise linear (or logistic) growth curve trend: Prophet automatically detects changes in trends by selecting changepoints from the data,
• a yearly seasonal component modeled using Fourier series,
• a weekly seasonal component using dummy variables,
• a user-provided list of important holidays.

In practice, non-linear trends are fit with yearly and weekly seasonality (plus holidays). The method is also robust to missing data, shifts in the trend, and large outliers.

Usage

Facebook has open sourced Prophet software, a forecasting project with an interface available in Python. We use this resource.

Run the tour_forecast.py source code or explore the run_forecast.ipynb notebook to produce the following 5-years forecast estimates of Eurostat tour_occ_nim monthly indicator:

Another example is provided by the 1-year prediction of unemployment une_rt_m monthly indicator:

About

status	since 2017 – closed
contributors
license	EUPL

Reference

• Taylor, S.J. and Letham, B. (2017): Forecasting at Scale.