package sklearn

scikit-learn for OCaml

ocaml-sklearn allows using Python's scikit-learn machine learning library from OCaml.

Read the online scikit-learn OCaml API documentation here.

If you are not familiar with scikit-learn, consult its Python getting started documentation and user guide.

This is a preview. The current OCaml API is not complete. Some functions may be hard or impossible to use. Also, the existing API is not stable, it may change to accomodate more functionality or make things easier to use.

Example : support vector regression with RBF kernel

let n_samples, n_features = 10, 5 in
Random.init 0;
let y = Sklearn.Ndarray.of_bigarray @@ Owl.Arr.uniform [|n_samples|] in
let x = Sklearn.Ndarray.of_bigarray @@ Owl.Dense.Matrix.D.uniform n_samples n_features in
let open Sklearn.Svm in
let clf = SVR.create ~c:1.0 ~epsilon:0.2 () in
Format.printf "%a\n" SVR.pp @@ SVR.fit clf ~x ~y;
Format.printf "%a\n" Sklearn.Arr.pp @@ SVR.support_vectors_ clf;;

This outputs:

SVR(C=1.0, cache_size=200, coef0=0.0, degree=3, epsilon=0.2, gamma='scale',
    kernel='rbf', max_iter=-1, shrinking=True, tol=0.001, verbose=False)
[[0.14509922 0.16277752 0.99033894 0.84013554 0.96508279]
 [0.8865312  0.80655193 0.07459775 0.36058768 0.22130337]
 [0.21844203 0.09612442 0.49908686 0.1154579  0.98202969]
 [0.07306658 0.97225754 0.20558949 0.16423512 0.57400651]
 [0.08153976 0.41462111 0.66190418 0.70208221 0.3600998 ]
 [0.20502873 0.04244781 0.21800856 0.28184598 0.4282653 ]
 [0.89211037 0.51466381 0.23432621 0.29850877 0.13323457]]

There are more examples in examples/auto, for instance examples/auto/svm.ml.

Installation

opam install sklearn

Finding Python's scikit-learn at runtime

At runtime, ocaml-sklearn expects to load the right version of Python's scikit-learn. One way to do that is to create a virtualenv, install scikit-learn version %%SKLEARN_FULL_VERSION%% inside, and run your OCaml program in the activated virtualenv.

Do this once to create the virtualenv in .venv and install scikit-learn inside:

python3 -mvenv .venv
source .venv/bin/activate
pip install scikit-learn==%%SKLEARN_FULL_VERSION%% pytest

Then run your compiled OCaml program inside the virtualenv:

source .venv/bin/activate
./my_ocaml_program.exe

A version of ocaml-sklearn is tied to a version of Python's sklearn. For instance, a version of ocaml-sklearn for Python's scikit-learn 0.22.2 will refuse to initialize (by throwing an exception) if scikit-learn's version is not 0.22 (it can be 0.22.1, 0.22.2 or 0.22.2.post1).

API

We attempt to bind all of scikit-learn's APIs. However, not all of the APIs are currently tested, and some are probably hard to use or unusable at the moment.

Each Python module or class gets its own OCaml module. For instance Python class sklearn.svm.SVC can be found in OCaml module Sklearn.Svm.SVC. This module has a create function to construct an SVC and functions corresponding to the Python methods and attributes.

Most data is passed in and out of sklearn through module Arr. An Arr.t is either a dense Ndarray.t or a sparse Csr_matrix.t.

You should generally build a dense array using the constructors in Arr:

let x = Arr.Float.matrix [|[| 1; 2 |]; [| 3; 4 |]|]

One way to build an Arr.t is to use Owl's function to construct a bigarray and then use Arr.of_bigarray. Data is shared between the bigarray and the Arr.t.

To get data out of an Arr.t, get the underlying Ndarray or Csr_matrix using Arr.get.

Attributes are exposed read-only, each with two getters: one that raises Not_found if the attribute is None, and the other that returns an option.

Bunches (as returned from the sklearn.datasets APIs) are exposed as objects.

Arguments taking string values are converted (in most cases) to polymorphic variants.

Each module has a conversion function to Py.Object.t, so that you can always escape and use pyml directly if the API provided here is incomplete.

No attempt is made to expose features marked as deprecated.

Development notes

ocaml-sklearn's sources are generated using a Python program (see lib/skdoc.py) that loads up sklearn and uses introspection to generate bindings based on pyml. To determine types, it parses scikit-learn's documentation.

Python requirements

The requirements for developing (not using) the bindings are in file requirements-dev.txt. Install it using:

python3 -mvenv .venv
source .venv/bin/activate
pip install -r requirements-dev.txt

Running tests

dune runtest

The tests are in examples/auto. They are based on examples extracted from the Python documentation. A good way to develop is to pick one of the files and start porting examples. One can refer to examples/auto/svm.ml or examples/auto/pipeline.ml, whose examples have already been ported (almost) completely.

The following examples have been ported completely:

• neighbors

• pipeline

• preprocessing

• svm

• ensemble

The following examples still need to be ported:

• calibration

• cluster

• compose

• covariance

• cross_decomposition

• datasets

• decomposition

• discriminant_analysis

• feature_extraction

• feature_selection

• gaussian_process

• impute

• inspection

• isotonic

• kernel_approximation

• linear_model

• manifold

• metrics

• model_selection

• multiclass

• multioutput

• naive_bayes

• neural_network

• random_projection

• semi_supervised

• tree

• utils

Generating documentation

lib/build-doc

Documentation can then be found in html_doc/. Serve it locally with something like:

python3 -mhttp.server --directory html_doc
xdg-open http://localhost:8000

License

BSD-3. See file LICENSE.