def test_forest_reg_model(data, test):
import numpy as np
from mle_training.utils import data_preprocess as preprocess
housing = data
# Fit missing value imputer on train data
preprocess.fit(train_data=housing)
# Transform train and test data
X_train, y_train = preprocess.transform(data=housing)
# Fit model and score on training set
forest_model = train_score.forest_reg_model(X=X_train, y=y_train)
y_pred_model = forest_model.predict(test)[0]
assert np.round(y_pred_model, 3) == 134796.667
def test_model_predict(data):
import numpy as np
from mle_training.utils import data_preprocess as preprocess
housing = data
# Fit missing value imputer on train data
preprocess.fit(train_data=housing)
# Transform train and test data
X_train, y_train = preprocess.transform(data=housing)
from mle_training import train_score # train and score module
# Fit model and score on training set
lin_model = train_score.linear_reg_model(X=X_train, y=y_train)
# Score trained model on test set (can be a model stored in a pickle file)
y_pred = predict.model_predict(model=lin_model, X=[X_train.iloc[0, :]])[0]
assert np.round(y_pred, 3) == 136237.050