bfast
bfast copied to clipboard
bfast2
Breaks For Additive Season and Trend
By default `bfastts` will fail with a cryptic message if no `type` is supplied. It should either give a reasonable message or assume `irregular` (or `regular` if #2 is implemented).
If we have the leap day (Feb 29) and also the day after (March 1) in the time series, `bfastts` fails with an error: ``` Error in merge.zoo(zoo(coredata(x), tt), zoo(,...
This call to `Fstat` in `bfast01()` causes an error for some inputs: https://github.com/GreatEmerald/bfast/blob/6e86571601510ea33af4daa16b0e36564995bd54/R/bfast01.R#L156 The error is `inadmissable change points: 'from' is larger than 'to'`. Setting `trim` to a low number...
Right now, running `fitted()` on a `bfast` object returns `NULL` (that's returned for any unknown class). It would be better if it returned `x$output[[iteration]]$St + x$output[[iteration]]$Tt` instead.
The current plot method in `plot.bfast` shows the components one by one, but there is no way to plot everything in one plot. It would be very convenient to have...
`zoo` is able to convert regular time series `zoo` objects into `ts` objects, so it is possible to automate making a `bfastts` from a `zoo` object that way. The problem...
This is not a coding question but more of a theoretical one, so if it doesn't belong here feel free to close it. I am about to use the [BFAST...
