Modelkit.Standard_scalerColumn-wise centering and population-standard-deviation scaling.
Fitting and transformation require finite inputs. Constant features use a scale of one, so centering maps them to zero without division by zero. Optional sample weights give weighted means and weighted population variances over positively weighted rows; an all-zero weight vector is a typed error. Fit and transform are O(rows * columns) and transform allocates one dense output matrix.
val create : ?with_mean:bool -> ?with_std:bool -> unit -> tinclude TRANSFORMER
with type t := t
and type params := params
and type target = unit
and type fitted := fitted
and type rng = Rng.ttype rng = Rng.tval fit :
t ->
?sample_weight:Sample_weight.t ->
rng:rng ->
feature_schema:Feature_schema.t ->
x:Matrix.t ->
y:target option ->
unit ->
(fitted, Error.t) resultval input_schema : fitted -> Feature_schema.tval output_schema : fitted -> Feature_schema.t