- Gaussian processes (http://www.gaussianprocess.org/gpml/chapters/RW2.pdf)
- Quantile regression forest
- Parameterize the moving window
- Track weights history in
backtest_regression()
- Multivariate TimeSeries
- Support categorical time series
The following could be useful TimeSeries:
- Time of day, day of week, week of month, month of year, etc
- Indicator variable for holidays (1 during holidays 0 else) in many regions
- Indicator for quarters
- Correlation coefficient
- More preprocessing: moving average, filters, ...
- A function such as
checkresiduals()used here https://otexts.com/fpp2/regression-evaluation.html
- seasonal plots (incl polar)
- scatterplots (for multivariate) and scatterplot matrix
- Collection of datasets
- Hierarchical reconciliation