def test_fit_predict(self, resource_loader):
"""Tests prediction after a fit"""
config = ModelConfig(
**{
"model_type": "text",
"example_type": QUERY_EXAMPLE_TYPE,
"label_type": CLASS_LABEL_TYPE,
"model_settings": {
"classifier_type": "logreg"
},
"params": {
"fit_intercept": True,
"C": 100
},
"features": {
"bag-of-words": {
"lengths": [1]
},
"freq": {
"bins": 5
},
"length": {},
},
})
model = TextModel(config)
examples = self.labeled_data.queries()
labels = self.labeled_data.intents()
model.initialize_resources(resource_loader, examples, labels)
model.fit(examples, labels)
assert model.predict([markup.load_query("hi").query]) == "greet"
assert model.predict([markup.load_query("bye").query]) == "exit"
def test_fit_predict(self, resource_loader):
"""Tests prediction after a fit"""
config = ModelConfig(
**{
'model_type': 'text',
'example_type': QUERY_EXAMPLE_TYPE,
'label_type': CLASS_LABEL_TYPE,
'model_settings': {
'classifier_type': 'logreg'
},
'params': {
'fit_intercept': True,
'C': 100
},
'features': {
'bag-of-words': {
'lengths': [1]
},
'freq': {
'bins': 5
},
'length': {}
}
})
model = TextModel(config)
examples = [q.query for q in self.labeled_data]
labels = [q.intent for q in self.labeled_data]
model.initialize_resources(resource_loader, examples, labels)
model.fit(examples, labels)
assert model.predict([markup.load_query('hi').query]) == 'greet'
assert model.predict([markup.load_query('bye').query]) == 'exit'
def test_extract_features(self, resource_loader):
"""Tests extracted features after a fit"""
config = ModelConfig(
**{
"model_type": "text",
"example_type": QUERY_EXAMPLE_TYPE,
"label_type": CLASS_LABEL_TYPE,
"model_settings": {"classifier_type": "logreg"},
"params": {"fit_intercept": True, "C": 100},
"features": {
"bag-of-words": {"lengths": [1]},
},
}
)
model = TextModel(config)
examples = self.labeled_data.queries()
labels = self.labeled_data.intents()
model.initialize_resources(resource_loader, examples, labels)
model.fit(examples, labels)
expected_features = {
"bag_of_words|length:1|ngram:hi": 1,
"bag_of_words|length:1|ngram:OOV": 1,
}
extracted_features = model.view_extracted_features(
markup.load_query("hi there").query
)
assert extracted_features == expected_features
def test_extract_features(self, resource_loader):
"""Tests extracted features after a fit"""
config = ModelConfig(
**{
'model_type': 'text',
'example_type': QUERY_EXAMPLE_TYPE,
'label_type': CLASS_LABEL_TYPE,
'model_settings': {
'classifier_type': 'logreg'
},
'params': {
'fit_intercept': True,
'C': 100
},
'features': {
'bag-of-words': {
'lengths': [1]
},
}
})
model = TextModel(config)
examples = [q.query for q in self.labeled_data]
labels = [q.intent for q in self.labeled_data]
model.initialize_resources(resource_loader, examples, labels)
model.fit(examples, labels)
expected_features = {
'bag_of_words|length:1|ngram:hi': 1,
'bag_of_words|length:1|ngram:there': 1
}
extracted_features = model.view_extracted_features(
markup.load_query('hi there').query)
assert extracted_features == expected_features
def test_fit_cv(self, resource_loader):
"""Tests fitting with param selection"""
config = ModelConfig(
**{
"model_type": "text",
"example_type": QUERY_EXAMPLE_TYPE,
"label_type": CLASS_LABEL_TYPE,
"model_settings": {"classifier_type": "logreg"},
"param_selection": {
"type": "k-fold",
"k": 10,
"grid": {"C": [10, 100, 1000], "fit_intercept": [True, False]},
},
"features": {
"bag-of-words": {"lengths": [1]},
"freq": {"bins": 5},
"length": {},
},
}
)
model = TextModel(config)
examples = self.labeled_data.queries()
labels = self.labeled_data.intents()
model.initialize_resources(resource_loader, examples, labels)
model.fit(examples, labels)
assert model._current_params
def test_fit(self, resource_loader):
"""Tests that a basic fit succeeds"""
config = ModelConfig(
**{
"model_type": "text",
"example_type": QUERY_EXAMPLE_TYPE,
"label_type": CLASS_LABEL_TYPE,
"model_settings": {
"classifier_type": "logreg"
},
"params": {
"fit_intercept": True,
"C": 100
},
"features": {
"bag-of-words": {
"lengths": [1]
},
"freq": {
"bins": 5
},
"length": {},
},
})
model = TextModel(config)
examples = self.labeled_data.queries()
labels = self.labeled_data.intents()
model.initialize_resources(resource_loader, examples, labels)
model.fit(examples, labels)
assert model._current_params == {"fit_intercept": True, "C": 100}
def test_fit(self, resource_loader):
"""Tests that a basic fit succeeds"""
config = ModelConfig(
**{
'model_type': 'text',
'example_type': QUERY_EXAMPLE_TYPE,
'label_type': CLASS_LABEL_TYPE,
'model_settings': {
'classifier_type': 'logreg'
},
'params': {
'fit_intercept': True,
'C': 100
},
'features': {
'bag-of-words': {
'lengths': [1]
},
'freq': {
'bins': 5
},
'length': {}
}
})
model = TextModel(config)
examples = [q.query for q in self.labeled_data]
labels = [q.intent for q in self.labeled_data]
model.initialize_resources(resource_loader, examples, labels)
model.fit(examples, labels)
assert model._current_params == {'fit_intercept': True, 'C': 100}
def test_fit_cv(self, resource_loader):
"""Tests fitting with param selection"""
config = ModelConfig(
**{
'model_type': 'text',
'example_type': QUERY_EXAMPLE_TYPE,
'label_type': CLASS_LABEL_TYPE,
'model_settings': {
'classifier_type': 'logreg'
},
'param_selection': {
'type': 'k-fold',
'k': 10,
'grid': {
'C': [10, 100, 1000],
'fit_intercept': [True, False]
},
},
'features': {
'bag-of-words': {
'lengths': [1]
},
'freq': {
'bins': 5
},
'length': {}
}
})
model = TextModel(config)
examples = [q.query for q in self.labeled_data]
labels = [q.intent for q in self.labeled_data]
model.initialize_resources(resource_loader, examples, labels)
model.fit(examples, labels)
assert model._current_params