Hello. I am Orkun Berk Yüzbasioglu. Currently, I am working as a Forecasting Specialist at the Business Analytics Department. I am interested in mainly geospatial statistics and forecasting.
The video is about the fable package which is a replacement for the famous Cran certified forecast package. forecast package is for single/univariate series, but fable can predict many/multivariate related time series and build on top of tsibble package. Also all of the fable models produce mabla class object, so there is more conistency.
This post is introducing the reader to time series modeling and its related techniques in R. In the post, a real world forecasting case is solved/explained step by step. A builtin dataset from R, namely AirPassangers is used, first this series is visualised. Then the stationary assumption is checked with Dickey-Fuller test, which must hold to apply both AR and MA techniques. Then diferencing and log is applied to the series for the stationary assumption to hold. In the last step, ARIMA was used to model the data and produce forecasts.
This article at r-bloggers.com is covering high level conceps and provides both the data and code for performing time series forecasting in R.
This post is about time series analysis and forecasting at R. Building a time series from a numerical data is explained. Also examples of building exponential smoothing models at base R and building ARIMA and exponential models at forecast package are mentioned in this post.