Merge pull request #29 from tqtg/master

Add tests for base_strategy and ratio

cornac/eval_strategies/ratio_split.py

@@ -12,7 +12,6 @@
 
 
 class RatioSplit(BaseStrategy):
-
     """Train-Test Split Evaluation Strategy.
 
     Parameters
@@ -48,15 +47,15 @@ class RatioSplit(BaseStrategy):
         Output running log
     """
 
-    def __init__(self, data, data_format='UIR', val_size=0.0, test_size=0.2, rating_threshold=1., shuffle=True, random_state=None,
-                 exclude_unknowns=False, verbose=False):
-        BaseStrategy.__init__(self, data = data, data_format='UIR', rating_threshold=rating_threshold, exclude_unknowns=exclude_unknowns, verbose=verbose)
+    def __init__(self, data, data_format='UIR', val_size=0.0, test_size=0.2, rating_threshold=1., shuffle=True,
+                 random_state=None, exclude_unknowns=False, verbose=False):
+        BaseStrategy.__init__(self, data=data, data_format=data_format, rating_threshold=rating_threshold,
+                              exclude_unknowns=exclude_unknowns, verbose=verbose)
 
         self._shuffle = shuffle
         self._random_state = random_state
         self._train_size, self._val_size, self._test_size = self._validate_sizes(val_size, test_size, len(self._data))
-        self._split_run = False
-
+        self._split_ran = False
 
 
     @staticmethod
@@ -93,6 +92,11 @@ def _validate_sizes(val_size, test_size, num_ratings):
 
 
     def split(self):
+        if self._split_ran:
+            if self.verbose:
+                print('Data is already split!')
+            return
+
         if self.verbose:
             print("Splitting the data")
 
@@ -114,15 +118,13 @@ def split(self):
         if self._data_format == 'UIR':
             self.build_from_uir_format(train_data, val_data, test_data)
 
-        self._split_run = True
+        self._split_ran = True
 
         if self.verbose:
             print('Total users = {}'.format(self.total_users))
             print('Total items = {}'.format(self.total_items))
 
 
     def evaluate(self, model, metrics, user_based):
-        if not self._split_run:
-            self.split()
-
+        self.split()
         return BaseStrategy.evaluate(self, model, metrics, user_based)

cornac/eval_strategies/tests/test_base_strategy.py

@@ -0,0 +1,34 @@
+# -*- coding: utf-8 -*-
+
+"""
+@author: Quoc-Tuan Truong <tuantq.vnu@gmail.com>
+"""
+
+from ..base_strategy import BaseStrategy
+
+
+def test_init():
+    bs = BaseStrategy(None, verbose=True)
+
+    assert not bs.exclude_unknowns
+    assert 1. == bs.rating_threshold
+
+
+def test_trainset_none():
+    bs = BaseStrategy(None, verbose=True)
+
+    try:
+        bs.evaluate(None, {}, False)
+    except ValueError:
+        assert True
+
+
+def test_testset_none():
+    bs = BaseStrategy(None, train_set=[], verbose=True)
+
+    try:
+        bs.evaluate(None, {}, False)
+    except ValueError:
+        assert True
+
+

cornac/eval_strategies/tests/test_ratio_split.py

@@ -0,0 +1,49 @@
+# -*- coding: utf-8 -*-
+
+"""
+@author: Quoc-Tuan Truong <tuantq.vnu@gmail.com>
+"""
+
+from ..ratio_split import RatioSplit
+
+
+def test_validate_size():
+    train_size, val_size, test_size = RatioSplit._validate_sizes(0.1, 0.2, 10)
+    assert 7 == train_size
+    assert 1 == val_size
+    assert 2 == test_size
+
+    train_size, val_size, test_size = RatioSplit._validate_sizes(None, 0.5, 10)
+    assert 5 == train_size
+    assert 0 == val_size
+    assert 5 == test_size
+
+    train_size, val_size, test_size = RatioSplit._validate_sizes(None, None, 10)
+    assert 10 == train_size
+    assert 0 == val_size
+    assert 0 == test_size
+
+    train_size, val_size, test_size = RatioSplit._validate_sizes(2, 2, 10)
+    assert 6 == train_size
+    assert 2 == val_size
+    assert 2 == test_size
+
+    try:
+        RatioSplit._validate_sizes(-1, 0.2, 10)
+    except ValueError:
+        assert True
+
+    try:
+        RatioSplit._validate_sizes(11, 0.2, 10)
+    except ValueError:
+        assert True
+
+    try:
+        RatioSplit._validate_sizes(0, 11, 10)
+    except ValueError:
+        assert True
+
+    try:
+        RatioSplit._validate_sizes(3, 8, 10)
+    except ValueError:
+        assert True

cornac/experiment/tests/test_experiment.py

@@ -13,7 +13,7 @@ def test_with_ratio_split():
     from ..experiment import Experiment
 
     data = reader.txt_to_uir_triplets('./cornac/data/tests/data.txt')
-    exp = Experiment(eval_strategy=RatioSplit(data),
+    exp = Experiment(eval_strategy=RatioSplit(data, verbose=True),
                      models=[PMF(1, 0)],
                      metrics=[MAE(), RMSE(), Recall(1), FMeasure(1)],
                      verbose=True)

cornac/utils/generic_utils.py

@@ -87,6 +87,12 @@ def safe_indexing(X, indices):
 
 
 def validate_data_format(data_format):
+    """Check the input data format is supported or not
+        - UIR: (user, item, rating) triplet data
+        - UIRT: (user, item , rating, timestamp) quadruplet data
+
+    :raise ValueError if not supported
+    """
     data_format = str(data_format).upper()
     if not data_format in ['UIR', 'UIRT']:
         raise ValueError('{} data format is not supported!'.format(data_format))