Rob Hyndman
@robjhyndman.com
Professor of Statistics, Monash University, Australia. FAA, FASSA. Interested in #forecasting, #rstats, #statistics. he/him http://robjhyndman.com
Code to produce the hex-sticker wall is provided here: mitchelloharawild.com/blog/user-20...
Mitchell O’Hara-Wild - useR! 2018 feature wall
Creating the hexmap feature wall for useR! 2018
mitchelloharawild.com
September 29, 2025 at 7:52 AM
Code to produce the hex-sticker wall is provided here: mitchelloharawild.com/blog/user-20...
The gratis package will simulate from several time series models, with either specified parameters or random parameters. It can also generate series that look "similar" to existing series. cran.r-project.org/package=gratis
gratis: Generating Time Series with Diverse and Controllable Characteristics
Generates synthetic time series based on various univariate time series models including MAR and ARIMA processes. Kang, Y., Hyndman, R.J., Li, F.(2020) <<a href="https://doi.org/10.1002%2Fsam.11461...
cran.r-project.org
August 10, 2025 at 10:14 PM
The gratis package will simulate from several time series models, with either specified parameters or random parameters. It can also generate series that look "similar" to existing series. cran.r-project.org/package=gratis
NNs will never win if used as a local model (fitted to a single series). But they can be better as a global model (fitted to many series). However, LightGBM has won recent forecasting competitions, not NNs.
July 6, 2025 at 5:46 AM
NNs will never win if used as a local model (fitted to a single series). But they can be better as a global model (fitted to many series). However, LightGBM has won recent forecasting competitions, not NNs.
Interesting. I haven't looked at state space models for more than 10 years. Moved on to other things.
May 1, 2025 at 12:00 AM
Interesting. I haven't looked at state space models for more than 10 years. Moved on to other things.