def model_train(model_id, features, target):
result = {}
model_path = os.path.join(config.model_dir, model_id + '.joblib')
model = ml_models.load_model(model_path)
#x_train, x_test, y_train, y_test = train_test_split(features, target)
x_train, y_train = features, target
x_test, y_test = features, target
model.fit(x_train, y_train)
y_pred = model.predict(x_test)
result["mean_squared_error"] = metrics.mean_squared_error(y_test, y_pred)
result["mean_absolute_error"] = metrics.mean_absolute_error(y_test, y_pred)
try:
train_loss = model.named_steps['model'].lpath['train']
val_loss = model.named_steps['model'].lpath['val']
except Exception:
train_loss = []
val_loss = []
result["train_loss"] = train_loss
result["val_loss"] = val_loss
ml_models.save_model(model, model_path)
return result
def model_create(model_id, model_type, parameters):
model_path = os.path.join(config.model_dir, model_id + '.joblib')
# get definition validator
validatorDefinition = ml_models.get_validator(model_type)
# fit value
validator = validatorDefinition(**parameters)
# validate
params = validator()
model = ml_models.build_model(model_type, params)
ml_models.save_model(model, model_path)
def model_train(model_id, features, target):
result = {}
model_path = os.path.join(config.model_dir, model_id + '.joblib')
model = ml_models.load_model(model_path)
#x_train, x_test, y_train, y_test = train_test_split(features, target)
x_train, y_train = features, target
x_test, y_test = features, target
model.fit(x_train, y_train)
y_pred = model.predict(x_test)
result["mean_squared_error"] = metrics.mean_squared_error(y_test, y_pred)
result["mean_absolute_error"] = metrics.mean_absolute_error(y_test, y_pred)
ml_models.save_model(model, model_path)
return result
def model_train(model_id, features, target, **kwags):
result = {}
model_path = os.path.join(config.model_dir, model_id+'.joblib')
model = ml_models.load_model(model_path)
x_train = features
y_train = target
model.fit(x_train, y_train, **kwags)
y_pred = model.predict(x_train)
cm = metrics.confusion_matrix(y_train, y_pred)
true_pred = 0
total = 0
for i in range (len(cm)):
true_pred += cm[i][i]
total += sum(cm[i])
result["accuracy"] = round(true_pred / total, 4)
ml_models.save_model(model, model_path)
return result