Discretizers.DiscretizeUniformWidth
— TypeDiscretizeUniformWidth(alg::Symbol) If nbins
is a symbol, automatically determine the number of bins to use
Discretizers.LinearDiscretizer
— TypeLinearDiscretizer
A discretizer which encodes (typically) continuous values into discrete bins. A univariate domain is divided into discrete by edges. A value V will be encoded into bin B if V ∈ [Bₗ Bᵣ) (or V ∈ [Bₗ Bᵣ] if B is the rightmost bin)
If forceoutliersto_closest is set to true, then values outside of binedges will be shunted into the nearest bin. Otherwise an error is thrown.
TODO(tim): handle NA handle Nullable handle NaN ability to specify decoding behavior
Discretizers.datalineardiscretizer
— Methoda hybrid discretizers that maps a special missing value indicator (typically Inf or NaN) to a discrete bin, but otherwise is a linear discretizer
Discretizers.decode
— MethodThere are several methods for decoding a LinearDiscretizer. The default is SampleUniform, which uniformly samples from the bin. SampleBinCenter returns the bin's center value.