Ndarray.t represents a tensor (a multidimensional array), which can contain integers, floats and strings.
You can **build an Ndarray from an OCaml array** :
```ocaml let nda = Sklearn.Ndarray.Float.matrix | [|1;2;3|]; [|4;5;6|] | in Format.printf "%a" Ndarray.pp nda;; ```
You can also **build it from a bigarray**. This way, no data is copied, the bigarray and the Ndarray share the same memory. You may want to use the library Owl to build the bigarray (this is useful because the interface provided here is minimal).
```ocaml let module Matrix = Owl.Dense.Matrix.D in let x = Sklearn.Ndarray.of_bigarray @@ Matrix.uniform 10 3 in Format.printf "%a" Ndarray.pp x;; ```
Finally, this module provides functions for **building a Python list of Ndarrays**, which is useful for functions expecting a collection of arrays which do not have the same length (and therefore do not fit into one tensor).
```ocaml let y = Ndarray.String.vectors [|"hello"; "world"|]; [|"how"; "are"; "you"; "gentlemen"|] in
Format.printf "%a" Ndarray.pp y;;
```
val show : t -> string### show
Pretty-print an Ndarray into a string.
val pp : Stdlib.Format.formatter -> t -> unit### pp
Pretty-print the Ndarray.
val to_pyobject : t -> Py.Object.t### to_pyobject
Convert the Ndarray to a Py.Object.t.
val of_pyobject : Py.Object.t -> t### of_pyobject
Build an Ndarray from a Py.Object.t.
val of_bigarray : ('a, 'b, 'c) Stdlib.Bigarray.Genarray.t -> t### of_bigarray
Build an Ndarray from a bigarray.
val shape : t -> int array### shape
Shape (dimensions) of an Ndarray.
module Float : sig ... end## module Ndarray.Float
module Int : sig ... end## module Ndarray.Int
module String : sig ... end## module Ndarray.String