def test_local_seriesprocessing(self):
data = np.random.randint(0, 2, size=(1000, 3))
data = pd.DataFrame(data=data, columns=["col1", "col2", "col3"])
model = Model(
x_train=data,
target_field="col3",
test_split_percentage=0.5,
report_name="modelweights",
)
model.logistic_regression(
random_state=2, penalty="l2", model_name="l1", run=True
)
model.logistic_regression(
random_state=2, penalty="l2", model_name="l2", run=True
)
model.logistic_regression(
random_state=2, penalty="l2", model_name="l3", run=True
)
model.run_models(method="series")
self.assertTrue(len(model._models) == 3 and len(model._queued_models) == 0)
def test_unsupervisedcv(self):
data = np.random.randint(0, 2, size=(1000, 3))
data = pd.DataFrame(data=data, columns=["col1", "col2", "col3"])
model = Model(x_train=data, target_field="col3", test_split_percentage=0.2)
model.kmeans(cv=2, random_state=2, learning_curve=True)
self.assertTrue(True)
def test_interpretmodel_behaviour_dependence(self):
data = np.random.randint(0, 2, size=(1000, 3))
data = pd.DataFrame(data=data, columns=["col1", "col2", "col3"])
model = Model(x_train=data, target_field="col3", test_split_percentage=0.4)
model.logistic_regression(random_state=2, run=True)
model.log_reg.interpret_model_behavior(method="dependence", show=False)
self.assertTrue(True)
def test_model_dbscan(self):
data = [[1, 2], [2, 2], [2, 3], [8, 7], [8, 8], [25, 80]]
data = pd.DataFrame(data=data, columns=["col1", "col2"])
model = Model(x_train=data, split=False)
model.dbscan(eps=3, min_samples=2, run=True)
validate = model.dbscan_clusters is not None
self.assertTrue(validate)
def test_model_kmeans(self):
data = [[1, 2], [2, 2], [2, 3], [8, 7], [8, 8], [25, 80]]
data = pd.DataFrame(data=data, columns=["col1", "col2"])
model = Model(x_train=data, split=False)
model.kmeans(n_clusters=3, random_state=0, run=True)
validate = model.kmeans_clusters is not None
self.assertTrue(validate)
def test_plot_clusters3d(self):
data = [[1, 2, 3], [2, 2, 3], [2, 3, 4], [8, 7, 5], [8, 8, 5], [25, 80, 4]]
data = pd.DataFrame(data=data, columns=["col1", "col2", "col3"])
model = Model(x_train=data, split=False)
model.kmeans(n_clusters=3, random_state=0, run=True)
model.km.plot_clusters(dim=3)
self.assertTrue(True)
def test_model_gaussianmixture(self):
data = [[1, 2], [2, 2], [2, 3], [8, 7], [8, 8], [25, 80]]
data = pd.DataFrame(data=data, columns=["col1", "col2"])
model = Model(x_train=data, split=False)
model.gaussian_mixture_clustering(run=True)
validate = model.gm_cluster is not None
self.assertTrue(validate)
def test_model_meanshift(self):
data = [[1, 2], [2, 2], [2, 3], [8, 7], [8, 8], [25, 80]]
data = pd.DataFrame(data=data, columns=["col1", "col2"])
model = Model(x_train=data, split=False)
model.mean_shift(run=True)
validate = model.mshift is not None
self.assertTrue(validate)
def test_model_xgbc(self):
data = np.random.randint(0, 2, size=(1000, 3))
data = pd.DataFrame(data=data, columns=["col1", "col2", "col3"])
model = Model(x_train=data, target_field="col3")
model.xgboost_classification(run=True)
validate = model.xgb_cls is not None
self.assertTrue(validate)
def test_model_report_modelweights(self):
data = np.random.randint(0, 2, size=(1000, 3))
data = pd.DataFrame(data=data, columns=["col1", "col2", "col3"])
model = Model(x_train=data, target_field="col3", report_name="modelweights")
model.logistic_regression(random_state=2, penalty="l2", run=True)
model.log_reg.model_weights()
self.assertTrue(True)
def test_interpretmodel_predictions_all(self):
data = np.random.randint(0, 2, size=(1000, 3))
data = pd.DataFrame(data=data, columns=["col1", "col2", "col3"])
model = Model(x_train=data, target_field="col3", test_split_percentage=0.6)
model.logistic_regression(random_state=2, run=True)
model.log_reg.interpret_predictions(show=False)
self.assertTrue(True)
def test_model_dtregression(self):
data = np.random.randint(0, 2, size=(1000, 3))
data = pd.DataFrame(data=data, columns=["col1", "col2", "col3"])
model = Model(x_train=data, target_field="col3")
model.decision_tree_regression(random_state=2, run=True)
validate = model.dt_reg is not None
self.assertTrue(validate)
def test_model_linearsvc(self):
data = np.random.randint(0, 2, size=(1000, 3))
data = pd.DataFrame(data=data, columns=["col1", "col2", "col3"])
model = Model(x_train=data, target_field="col3")
model.linearsvc(random_state=2, run=True)
validate = model.linsvc is not None
self.assertTrue(validate)
def test_model_isoforest(self):
data = np.random.randint(0, 2, size=(1000, 3))
data = pd.DataFrame(data=data, columns=["col1", "col2", "col3"])
model = Model(x_train=data)
model.isolation_forest(random_state=2, run=True)
validate = model.iso_forest is not None
self.assertTrue(validate)
def test_model_oneclasssvm(self):
data = np.random.randint(0, 2, size=(1000, 3))
data = pd.DataFrame(data=data, columns=["col1", "col2", "col3"])
model = Model(x_train=data)
model.oneclass_svm(run=True)
validate = model.ocsvm is not None
self.assertTrue(validate)
def test_stratified_cv(self):
data = np.random.randint(0, 2, size=(1000, 3))
data = pd.DataFrame(data=data, columns=["col1", "col2", "col3"])
model = Model(x_train=data, target_field="col3", test_split_percentage=0.2)
cv_values = model.logistic_regression(
cv="strat-kfold", random_state=2, learning_curve=True, run=False
)
self.assertIsNotNone(len(cv_values) == 5)
def test_model_confusionmatrix(self):
data = np.random.randint(0, 2, size=(1000, 3))
data = pd.DataFrame(data=data, columns=["col1", "col2", "col3"])
model = Model(x_train=data, target_field="col3")
model.logistic_regression(random_state=2, penalty="l2", run=True)
model.log_reg.confusion_matrix()
self.assertTrue(True)
def test_model_agglom(self):
data = [[1, 2], [2, 2], [2, 3], [8, 7], [8, 8], [25, 80]]
data = pd.DataFrame(data=data, columns=["col1", "col2"])
model = Model(x_train=data, split=False)
model.agglomerative_clustering(n_clusters=2, run=True)
validate = model.agglom is not None
self.assertTrue(validate)
def test_model_svr(self):
data = np.random.randint(0, 2, size=(1000, 3))
data = pd.DataFrame(data=data, columns=["col1", "col2", "col3"])
model = Model(x_train=data, target_field="col3")
model.svr(run=True)
validate = model.svr_reg is not None
self.assertTrue(validate)
def test_dependence_plot(self):
data = np.random.randint(0, 2, size=(1000, 3))
data = pd.DataFrame(data=data, columns=["col1", "col2", "col3"])
model = Model(x_train=data, target_field="col3", test_split_percentage=0.5)
model.logistic_regression(random_state=2, penalty="l2", run=True)
model.log_reg.dependence_plot("col1")
self.assertTrue(True)
def test_force_plot_misclassified(self):
data = np.random.randint(0, 2, size=(1000, 3))
data = pd.DataFrame(data=data, columns=["col1", "col2", "col3"])
model = Model(x_train=data, target_field="col3", test_split_percentage=0.6)
model.logistic_regression(random_state=2, run=True)
model.log_reg.force_plot(misclassified=True)
self.assertTrue(True)
def test_model_cluster_filter(self):
data = [[1, 2], [2, 2], [2, 3], [8, 7], [8, 8], [25, 80]]
data = pd.DataFrame(data=data, columns=["col1", "col2"])
model = Model(x_train=data, split=False)
model = model.dbscan(eps=3, min_samples=2, run=True)
filtered = model.filter_cluster(0)
validate = all(filtered.dbscan_clusters == 0)
self.assertTrue(validate)
def test_model_unsupervised_defaultgridsearch(self):
data = np.random.randint(0, 2, size=(1000, 3))
data = pd.DataFrame(data=data, columns=["col1", "col2", "col3"])
model = Model(x_train=data, target_field="col3", report_name="gridsearch_test")
gridsearch_params = {"max_iter": [300, 200]}
model.kmeans(gridsearch=gridsearch_params, cv=2, run=True)
self.assertTrue(True)
def test_interpretmodel_behaviour_all(self):
train_data = np.random.random_sample(size=(1000, 2))
label_data = np.random.randint(0, 2, size=(1000, 1))
data = pd.DataFrame(data=train_data, columns=["col1", "col2"])
data["col3"] = label_data
model = Model(x_train=data, target_field="col3", test_split_percentage=0.2)
model.logistic_regression(random_state=2, run=True)
model.log_reg.interpret_model_behavior(show=False)
self.assertTrue(True)
def test_model_getattr(self):
text_data = [
"Hi my name is PyAutoML. Please split me.",
"This function is going to split by sentence. Automation is great.",
]
data = pd.DataFrame(data=text_data, columns=["data"])
model = Model(x_train=data, split=False)
model.extract_keywords_gensim("data", ratio=0.5, model_name="model1", run=True)
validate = model.model1 is not None and model["model1"] is not None
self.assertTrue(validate)
def test_text_gensim_summarize(self):
text_data = [
"Hi my name is PyAutoML. Please split me.",
"This function is going to split by sentence. Automation is great.",
]
data = pd.DataFrame(data=text_data, columns=["data"])
model = Model(x_train=data, split=False)
model.summarize_gensim("data", ratio=0.5, run=True)
validate = model.data_summarized is not None
self.assertTrue(validate)
def test_text_d2v(self):
text_data = [
"Hi my name is PyAutoML. Please split me.",
"This function is going to split by sentence. Automation is great.",
]
data = pd.DataFrame(data=text_data, columns=["data"])
model = Model(x_train=data, split=False)
model.doc2vec("data", prep=True, run=True, min_count=1)
validate = model.d2v is not None
self.assertTrue(validate)
def test_decision_plot_sameaxis(self):
data = np.random.randint(0, 2, size=(1000, 3))
data = pd.DataFrame(data=data, columns=["col1", "col2", "col3"])
model = Model(x_train=data, target_field="col3", test_split_percentage=0.5)
model.logistic_regression(random_state=2, penalty="l2", run=True)
r = model.log_reg.decision_plot(sample_no=1)
model.log_reg.decision_plot(
sample_no=2, feature_order=r.feature_idx, xlim=r.xlim
)
self.assertTrue(True)
def test_model_logisticregression(self):
data = np.random.randint(0, 2, size=(1000, 3))
data = pd.DataFrame(data=data, columns=["col1", "col2", "col3"])
model = Model(x_train=data, target_field="col3")
model.logistic_regression(random_state=2, penalty="l2", run=True)
validate = (
model.x_train_results.log_predictions is not None
and model.x_test_results.log_predictions is not None
)
self.assertTrue(validate)
def test_text_w2vprep(self):
text_data = [
"Hi my name is PyAutoML. Please split me.",
"This function is going to split by sentence. Automation is great.",
]
data = pd.DataFrame(data=text_data, columns=["data"])
data["prep"] = pd.Series([text.split() for text in text_data])
model = Model(x_train=data, split=False)
model.word2vec("prep", run=True, min_count=1)
validate = model.w2v is not None
self.assertTrue(validate)