Prophet模型使用for time-series时间序列模型

2018-12-19 本文已影响0人数据小新手

$y(t)=g(t)+s(t)+\epsilon_t$

g(t): piecewise linear or logistic growth curve for modelling non-periodic changes in time series
s(t): periodic changes (e.g. weekly/yearly seasonality)
h(t): effects of holidays (user provided) with irregular schedules
εt: error term accounts for any unusual changes not accommodated by the model

增长的最大值或缩减的最小值，可以设置cap来限制范围

可以设置关键点，比如新品发布，不可见的灾难，可以自己学习到也，可以手动设定

$s(t)=\sum(a_ncos(\frac{2\pi nt}{P})+b_nsin(\frac{2\pi nt}{P}))$

P is the period (365.25 for yearly data and 7 for weekly data)

Warning：

适用于连续的数据，每天或每小时都有数据，不太适合量化数据。因为交易时间有限
Reference: