def test_train_svm_model(self, model_type, c_val, kernel, expected):
# Create new Model
model_params = {}
model_params['C Parameter'] = c_val
model_params['Kernel Type'] = kernel
model = predict.make_model(model_type, model_params)
return_str = predict.train_model(model, test_run=True)
self.assertEqual(expected, return_str)
def test_train_knn_model(self, model_type, n_neighbors_val, algorithm_val,
expected):
# Create new Model
model_params = {}
model_params['Number of Neighbors'] = n_neighbors_val
model_params['Algorithm Type'] = algorithm_val
model = predict.make_model(model_type, model_params)
return_str = predict.train_model(model, test_run=True)
self.assertEqual(expected, return_str)
def test_train_rf_model(self, model_type, n_estimators_val, max_depth_val,
expected):
# Create new Model
model_params = {}
model_params['Number of Estimators'] = n_estimators_val
model_params['Max Depth'] = max_depth_val
model = predict.make_model(model_type, model_params)
return_str = predict.train_model(model, test_run=True)
self.assertEqual(expected, return_str)
def test_train_nn_model(self, model_type, solver_val, activation_val,
max_iter_val, expected):
# Create new Model
model_params = {}
model_params['Solver Type'] = solver_val
model_params['Activation Function Type'] = activation_val
model_params['Number of Iterations'] = max_iter_val
model = predict.make_model(model_type, model_params)
return_str = predict.train_model(model, test_run=True)
self.assertEqual(expected, return_str)
license_details = license_details.rename(
columns={'LicenseNumber': 'License #'})
#labeling and feature engineering
license_full_data = aggregate_permit_data(all_permits)
#finding contractors with ADU historical data
adu_builders = license_full_data[license_full_data['ADU_proj_count'] >= 1]
adu_builders = adu_builders.merge(license_details, on='License #', how='left')
#saving adu-builder data
adu_builders.to_csv('./outputs/adu_builders.csv')
adu_builders = pd.read_csv('./outputs/adu_builders.csv')
# finding non-adu contractors to predict on
non_adu_builders = license_full_data[
(license_full_data['ADU_proj_count'] < 2)
& (license_full_data['non_ADU_proj_count'] >= 2)]
# training the model on adu-contractors
trained_model = train_model(adu_builders)
# predicting for non-ADU contractors
ranked_non_adu_builders = predict_adu_performance_non_ADU_builders(
trained_model, non_adu_builders)
ranked_non_adu_builders = ranked_non_adu_builders.merge(license_details,
on='License #',
how='left')
ranked_non_adu_builders.to_csv('./outputs/ranked_contractors.csv')
topk = int(parse_results.topk) #--topk 5
cat_fileName = parse_results.cat_fileName #--cat_fileName cat_to_name.json
image_path = parse_results.image_path
#LoadData
data_transforms, image_datasets, dataloaders = predict.data_transforms()
#build the Model
model = predict.modelSetup(arch)
#froze paramter
predict.frozeParameters(model)
#Getting device info.
device = torch.device("cuda:0" if device == 'gpu' else "cpu")
#Build and trainning the classifer
classifier, model, criterion, optimizer = predict.make_classifier(
arch, model, device, hidden_units, dropout, learning_rate)
predict.train_model(dataloaders, optimizer, model, device, epochs, criterion)
#Save CheckPoint
predict.saveCheckPoint(model, image_datasets, epochs, optimizer)
#Accuracy
predict.validation_accuracy(dataloaders["test"], model, device, learning_rate)
print('Model Trained')
#Get Category Name
#--cat_fileName
#cat_to_name = predict.label_mapping(cat_fileName)