The function extracts features from functional data based on the DTW distance with a reference dataframe.

```
extractFDADTWKernel(
ref.method = "random",
n.refs = 0.05,
refs = NULL,
dtwwindow = 0.05
)
```

(data.frame).

- ref.method
(

`character(1)`

)

How should the reference curves be obtained? Method`random`

draws`n.refs`

random reference curves, while`all`

uses all curves as references. In order to use user-provided reference curves, this parameter is set to`fixed`

.- n.refs
(

`numeric(1)`

)

Number of reference curves to be drawn (as a fraction of the number of observations in the training data).- refs
(

`matrix`

|`integer(n)`

)

Integer vector of training set row indices or a matrix of reference curves with the same length as the functionals in the training data. Overwrites`ref.method`

and`n.refs`

.- dtwwindow
(

`numeric(1)`

)

Size of the warping window size (as a proportion of query length).

Other fda_featextractor:
`extractFDABsignal()`

,
`extractFDAFPCA()`

,
`extractFDAFourier()`

,
`extractFDAMultiResFeatures()`

,
`extractFDATsfeatures()`

,
`extractFDAWavelets()`