README.md

In kdwang1808/Dejavu: Deja vu: forecasting with similarity

Dejavu

The R package Dejavu provides efficient algorithms for forecasting with Similarity of a bunch of time series.

Installation

You can install Dejavu package from GitHub Repository with:

devtools::install_github("kdwang1808/Dejavu")

Usage

Load the package

require("Dejavu")

Forecasting with Similarity

• Using Reference data from M3 Competition (Download the repository to local first)

R library(dtw) library(robustbase) library(forecast) fc_Simi <- Similarity(AirPassengers, fh = 20, LoadData = TRUE, path = NULL) fcs_result <- ts(fc_Simi$fcs, start = 1961, frequency = 12) PIL_result <- ts(fc_Simi$PIL, start = 1961, frequency = 12) PIU_result <- ts(fc_Simi$PIU, start = 1961, frequency = 12) autoplot(AirPassengers)+autolayer(fcs_result)+autolayer(PIL_result)+autolayer(PIU_result)

• Using "Mydata" as Reference

R fc_Simi <- Similarity(AirPassengers, fh = 20, LoadData = FALSE, path = "Mydata")

References

• Yanfei Kang, Evangelos Spiliotis, Fotios Petropoulos, Nikolaos Athiniotis, Feng Li, Vassilios Assimakopoulos (2019). D´ej`a vu: forecasting with similarity.

Working paper on arXiv.

License

This package is free and open source software, licensed under GPL-3.

kdwang1808/Dejavu documentation built on March 23, 2020, 5:14 a.m.