/doc/api.rst

  1.. _api:
  2
  3API Documentation
  4=================
  5
  6.. _`embed model`:
  7
  8Embed the model
  9---------------
 10
 11One of the most common use of the mltool API is to embed the ranker model built
 12with the command line tool into your own application.
 13
 14The models are pickable file and you can load them using the :mod:`pickle` standard
 15module:
 16
 17.. code-block:: python
 18
 19    import pickle
 20    with open('model.pkl', 'rb') as fmodel:
 21        model = pickle.load(fmodel)
 22
 23Then you can use mltool API to predict the score of a sample:
 24
 25.. code-block:: python
 26
 27    from mltool.predict import predict
 28    pred = predict(model, {'f0': 1.0,
 29                           'f1': 0.0,
 30                           'f2': 0.0,
 31                           'f3': 0.2,
 32                           'f4': 1.0})
 33
 34Check :func:`predict <mltool.predict.predict>` and
 35:func:`predict_all <mltool.predict.predict_all>` functions for more details.
 36
 37.. module:: mltool
 38
 39Prediction
 40----------
 41
 42.. _predict:
 43
 44.. automodule:: mltool.predict
 45
 46-----------------
 47
 48.. autofunction:: predict
 49
 50-----------------
 51
 52.. autofunction:: predict_all
 53
 54.. _train:
 55
 56Model train
 57-----------
 58
 59mltool implements some algorithms to train regression models.
 60Currently it mainly supports Random Forest and regression trees.
 61
 62mltool.forest
 63~~~~~~~~~~~~~
 64
 65.. module:: mltool.forest
 66.. autofunction:: train_random_forest
 67
 68mltool.decisiontree
 69~~~~~~~~~~~~~~~~~~~
 70
 71.. module:: mltool.decisiontree
 72.. autofunction:: train_decision_tree
 73
 74
 75Model Evaluation
 76----------------
 77
 78.. _evaluate:
 79
 80.. automodule:: mltool.evaluate
 81
 82-----------------
 83
 84.. autofunction:: evaluate_preds
 85.. autofunction:: evaluate_model
 86
 87
 88.. _utilities:
 89
 90Utilities
 91---------
 92
 93.. module:: mltool.utils
 94
 95.. _dataset:
 96
 97Handling Dataset
 98~~~~~~~~~~~~~~~~
 99
100.. class:: Dataset
101
102    The Dataset class is a :func:`namedtuple <collections.namedtuple>` which
103    represents a set of samples with their labels and query ids.
104
105    .. attribute:: labels
106
107        An :class:`array <numpy.array>` of labels. Each label is a `float`.
108
109    .. attribute:: queries
110
111        A list of query ids.
112
113    .. attribute:: Dataset.samples
114
115        A 2d-:class:`array <numpy.array` of samples. It consists of one sample per
116        column, and one row for each feature.
117
118    .. attribute:: Dataset.feature_names
119
120        A sequence of feature names. Features are in the same order as they appear
121        in the :attr:`samples` rows.
122
123-----------------
124
125.. autofunction:: read_input_file
126
127