Facebook Prophet

Prophet is a time series forecasting tool open sourced by Facebook that aims to account for various seasonality of behavior as well as auto detecting change points and anticipating different behavior during holidays. Details can be found on the Prophet page.

Warning

This is still a work in progress. Currently the plugin shells out to a Python script that runs the Prophet python module. It’s less than efficient but it’s enough to get started building a UI.

Fields specific to the prophet config include:

Name

Data Type

Required

Description

Default

Example

numberofChangePoints

Numeric

Optional

The number of anticpated change points spread evenly across the historical data. Must be smaller than the number of points fed into the training algo.

25

1440

growth

String

Optional

One of LINEAR, LOGISTIC, FLAT. The anticipated change in values over time, e.g. growing linearly or logarithmically.

LINEAR

LOGISTIC

changePointRange

Numeric

Optional

Proportion of history in which trend changepoints will be estimated. Defaults to 0.8 for the first 80%. Not used if changepoints (not implemented yet) is specified.

0.80

0.50

yearlySeasonality

Boolean

Optional

Whether or not to look for yearly seasonality. If null, defaults to auto detection.

null

true

weeklySeasonality

Boolean

Optional

Whether or not to look for weekly seasonality. If null, defaults to auto detection.

null

true

dailySeasonality

Boolean

Optional

Whether or not to look for daily seasonality. If null, defaults to auto detection.

null

true

seasonalityMode

String

Optional

One of ADDITIVE or MULTIPLICATIVE. TODO not sure what it is.

ADDITIVE

MULTIPLICATIVE

seasonalityPriorScale

Numeric

Optional

Parameter modulating the strength of the seasonality model. Larger values allow the model to fit larger seasonal fluctuations, smaller values dampen the seasonality. Can be specified for individual seasonalities using add_seasonality.

10

25

holidayPriorScale

Numeric

Optional

Parameter modulating the strength of the holiday components model, unless overridden in the holidays input.

10

25

changepointPriorScale

Numeric

Optional

Parameter modulating the flexibility of the automatic changepoint selection. Large values will allow many changepoints, small values will allow few changepoints.

0.05

0.10

mcmcSamples

Numeric

Optional

If greater than 0, will do full Bayesian inference with the specified number of MCMC samples. If 0, will do MAP estimation

0

5

uncertaintyIntervalWidth

Numeric

Optional

Width of the uncertainty intervals provided for the forecast. If mcmcSamples = 0, this will be only the uncertainty in the trend using the MAP estimate of the extrapolated generative model. If mcmcSamples > 0, this will be integrated over all model parameters, which will include uncertainty in seasonality.

0.8

0.5

uncertaintySamples

Numeric

Optional

Number of simulated draws used to estimate uncertainty intervals. Settings this value to 0 or False will disable uncertainty estimation and speed up the calculation.

1000

0