Wrapper of `automap::autofitVariogram`

to facilitate computing
variograms for multiple genes in SFE objects as an EDA tool. These functions
are written to conform to a uniform format for univariate methods to be
called internally. These functions are not exported, but the documentation is
written to show users the extra arguments to use when alling
`calculateUnivariate`

or `runUnivariate`

.

## Usage

```
.variogram(x, coords_df, formula = x ~ 1, scale = TRUE, ...)
.variogram_bv(x, y, coords_df, scale = TRUE, map = FALSE, ...)
.cross_variogram(x, y, coords_df, scale = TRUE, ...)
.cross_variogram_map(x, y, coords_df, width, cutoff, scale = TRUE, ...)
.variogram_map(x, coords_df, formula = x ~ 1, width, cutoff, scale = TRUE, ...)
```

## Arguments

- x
A numeric vector whose variogram is computed.

- coords_df
A

`sf`

data frame with the geometry and regressors for variogram modeling.- formula
A formula defining the response vector and (possible) regressors, in case of absence of regressors, use x ~ 1.

- scale
Logical, whether to scale

`x`

. Defaults to`TRUE`

so the variogram is easier to interpret and is more comparable between features with different magnitudes when the length scale of spatial autocorrelation is of interest.- ...
Other arguments passed to

`automap::autofitVariogram`

such as`model`

and`variogram`

such as`alpha`

for anisotropy. Note that`gstat`

does not fit ansotropic models and you will get a warning if you specify`alpha`

. Nevertheless, plotting the empirical anisotropic variograms and comparing them to the variogram fitted to the entire dataset can be a useful EDA tool.- y
For bivariate, another numeric vector whose variogram is computed.

- map
logical; if TRUE, and

`cutoff`

and`width`

are given, a variogram map is returned. This requires package sp. Alternatively, a map can be passed, of class SpatialDataFrameGrid (see sp docs)- width
the width of subsequent distance intervals into which data point pairs are grouped for semivariance estimates

- cutoff
spatial separation distance up to which point pairs are included in semivariance estimates; as a default, the length of the diagonal of the box spanning the data is divided by three.