From 0b88d486322391d4f0cd4b9c3129936196d8adc2 Mon Sep 17 00:00:00 2001
From: Tiago Freitas Pereira <tiagofrepereira@gmail.com>
Date: Mon, 16 Mar 2020 17:43:24 +0100
Subject: [PATCH] Created functions to break down a sampleset in order to use
the estimator.fit properly and a function that transform samplesets
---
bob/pipelines/sample.py | 39 +++++++++++++++++++++++++++++++++++++++
1 file changed, 39 insertions(+)
diff --git a/bob/pipelines/sample.py b/bob/pipelines/sample.py
index b5da013..2fffa92 100644
--- a/bob/pipelines/sample.py
+++ b/bob/pipelines/sample.py
@@ -1,6 +1,45 @@
"""Base definition of sample"""
+def samplesets_to_samples(samplesets):
+ """
+ Given a list of :py:class:`SampleSet` break them in to a list of :py:class:`Sample` with its
+ corresponding key
+
+ This is supposed to fit the :py:meth:`sklearn.estimator.BaseEstimator.fit` where X and y are the inputs
+ Check here https://scikit-learn.org/stable/developers/develop.html for more info
+
+ Parameters
+ ----------
+ samplesets: list
+ List of :py:class:`SampleSet
+
+
+ Return
+ ------
+ X and y used in :py:meth:`sklearn.estimator.BaseEstimator.fit`
+
+ """
+
+ # TODO: Is there a way to make this operation more efficient? numpy.arrays?
+ X = []
+ y= []
+
+ for s in samplesets:
+ X += s.samples
+ y += [s.key]
+
+ return X, y
+
+
+def transform_sample_sets(transformer, sample_sets):
+ return [
+ SampleSet(transformer.transform(sset.samples), parent=sset)
+ for sset in sample_sets
+ ]
+
+
+
def _copy_attributes(s, d):
"""Copies attributes from a dictionary to self
"""
--
GitLab