def test_standard_scaler_deserializer(self):
extract_features = ['a']
feature_extractor = FeatureExtractor(input_scalars=['a'],
output_vector='extracted_a_output',
output_vector_items=["{}_out".format(x) for x in extract_features])
# Serialize a standard scaler to a bundle
standard_scaler = StandardScaler(with_mean=True,
with_std=True
)
standard_scaler.mlinit(prior_tf=feature_extractor,
output_features='a_scaled')
standard_scaler.fit(self.df[['a']])
standard_scaler.serialize_to_bundle(self.tmp_dir, standard_scaler.name)
# Now deserialize it back
node_name = "{}.node".format(standard_scaler.name)
standard_scaler_tf = StandardScaler()
standard_scaler_tf = standard_scaler_tf.deserialize_from_bundle(self.tmp_dir, node_name)
# Transform some sample data
res_a = standard_scaler.transform(self.df[['a']])
res_b = standard_scaler_tf.transform(self.df[['a']])
self.assertEqual(res_a[0], res_b[0])
self.assertEqual(standard_scaler.name, standard_scaler_tf.name)
self.assertEqual(standard_scaler.op, standard_scaler_tf.op)
self.assertEqual(standard_scaler.mean_, standard_scaler_tf.mean_)
self.assertEqual(standard_scaler.scale_, standard_scaler_tf.scale_)
def test_standard_scaler_serializer(self):
standard_scaler = StandardScaler(with_mean=True, with_std=True)
standard_scaler.mlinit(input_features='a', output_features='a_scaled')
standard_scaler.fit(self.df[['a']])
standard_scaler.serialize_to_bundle(self.tmp_dir, standard_scaler.name)
expected_mean = self.df.a.mean()
expected_std = np.sqrt(np.var(self.df.a))
expected_model = {
"op": "standard_scaler",
"attributes": {
"mean": {
"type": {
"type": "tensor",
"tensor": {
"base": "double"
}
},
"value": {
"values": [expected_mean],
"dimensions": [1]
}
},
"std": {
"type": {
"type": "tensor",
"tensor": {
"base": "double"
}
},
"value": {
"values": [expected_std],
"dimensions": [1]
}
}
}
}
self.assertAlmostEqual(expected_mean,
standard_scaler.mean_.tolist()[0],
places=7)
self.assertAlmostEqual(expected_std,
np.sqrt(standard_scaler.var_.tolist()[0]),
places=7)
# Test model.json
with open("{}/{}.node/model.json".format(
self.tmp_dir, standard_scaler.name)) as json_data:
model = json.load(json_data)
self.assertEqual(standard_scaler.op, expected_model['op'])
self.assertEqual(
expected_model['attributes']['mean']['value']['dimensions'][0],
model['attributes']['mean']['value']['dimensions'][0])
self.assertEqual(
expected_model['attributes']['std']['value']['dimensions'][0],
model['attributes']['std']['value']['dimensions'][0])
self.assertAlmostEqual(
expected_model['attributes']['mean']['value']['values'][0],
model['attributes']['mean']['value']['values'][0],
places=7)
self.assertAlmostEqual(
expected_model['attributes']['std']['value']['values'][0],
model['attributes']['std']['value']['values'][0],
places=7)
# Test node.json
with open("{}/{}.node/node.json".format(
self.tmp_dir, standard_scaler.name)) as json_data:
node = json.load(json_data)
self.assertEqual(standard_scaler.name, node['name'])
self.assertEqual(standard_scaler.input_features,
node['shape']['inputs'][0]['name'])
self.assertEqual(standard_scaler.output_features,
node['shape']['outputs'][0]['name'])
def test_standard_scaler_serializer(self):
standard_scaler = StandardScaler(with_mean=True, with_std=True)
extract_features = ['a']
feature_extractor = FeatureExtractor(
input_scalars=['a'],
output_vector='extracted_a_output',
output_vector_items=["{}_out".format(x) for x in extract_features])
standard_scaler.mlinit(prior_tf=feature_extractor,
output_features='a_scaled')
standard_scaler.fit(self.df[['a']])
standard_scaler.serialize_to_bundle(self.tmp_dir, standard_scaler.name)
expected_mean = self.df.a.mean()
expected_std = np.sqrt(np.var(self.df.a))
expected_model = {
"op": "standard_scaler",
"attributes": {
"mean": {
"double": [expected_mean],
"shape": {
"dimensions": [{
"size": 1,
"name": ""
}]
},
"type": "tensor"
},
"std": {
"double": [expected_std],
"shape": {
"dimensions": [{
"size": 1,
"name": ""
}]
},
"type": "tensor"
}
}
}
self.assertAlmostEqual(expected_mean,
standard_scaler.mean_.tolist()[0],
places=7)
self.assertAlmostEqual(expected_std,
np.sqrt(standard_scaler.var_.tolist()[0]),
places=7)
# Test model.json
with open("{}/{}.node/model.json".format(
self.tmp_dir, standard_scaler.name)) as json_data:
model = json.load(json_data)
self.assertEqual(standard_scaler.op, expected_model['op'])
self.assertEqual(
expected_model['attributes']['mean']['shape']['dimensions'][0]
['size'],
model['attributes']['mean']['shape']['dimensions'][0]['size'])
self.assertEqual(
expected_model['attributes']['std']['shape']['dimensions'][0]
['size'],
model['attributes']['std']['shape']['dimensions'][0]['size'])
self.assertAlmostEqual(
expected_model['attributes']['mean']['double'][0],
model['attributes']['mean']['double'][0],
places=7)
self.assertAlmostEqual(
expected_model['attributes']['std']['double'][0],
model['attributes']['std']['double'][0],
places=7)
# Test node.json
with open("{}/{}.node/node.json".format(
self.tmp_dir, standard_scaler.name)) as json_data:
node = json.load(json_data)
self.assertEqual(standard_scaler.name, node['name'])
self.assertEqual(standard_scaler.input_features,
node['shape']['inputs'][0]['name'])
self.assertEqual(standard_scaler.output_features,
node['shape']['outputs'][0]['name'])
