def test_regressor():
X = [[0]] * 4 # ignored
y = [1, 2, 1, 1]
reg = DummyRegressor()
reg.fit(X, y)
assert_array_equal(reg.predict(X), [5. / 4] * len(X))
def mean_model(features, solutions, verbose=0):
columns = solutions.columns
clf = DummyRegressor()
print('Training Model... ')
clf.fit(features, solutions)
print('Done Training')
return (clf, columns)
def test_y_mean_attribute_regressor():
X = [[0]] * 5
y = [1, 2, 4, 6, 8]
# when strategy = 'mean'
est = DummyRegressor(strategy='mean')
est.fit(X, y)
assert_equal(est.y_mean_, np.mean(y))
def train_classifier(): X_train = tfv.transform(video_captions_train) X_test = tfv.transform(video_captions_test) dummy = DummyRegressor(strategy="median") dummy.fit(X_train, Y_train) Y_pred_med = dummy.predict(X_test)
def test_dummy_regressor_on_nan_value():
X = [[np.NaN]]
y = [1]
y_expected = [1]
clf = DummyRegressor()
clf.fit(X, y)
y_pred = clf.predict(X)
assert_array_equal(y_pred, y_expected)
def test_dummy_regressor_on_3D_array():
X = np.array([[['foo']], [['bar']], [['baz']]])
y = np.array([2, 2, 2])
y_expected = np.array([2, 2, 2])
cls = DummyRegressor()
cls.fit(X, y)
y_pred = cls.predict(X)
assert_array_equal(y_pred, y_expected)
class Regressor(BaseEstimator):
def __init__(self):
self.clf = DummyRegressor()
def fit(self, X, y):
self.clf.fit(X, y)
def predict(self, X):
return self.clf.predict(X)
def test_scorer_sample_weight():
# Test that scorers support sample_weight or raise sensible errors
# Unlike the metrics invariance test, in the scorer case it's harder
# to ensure that, on the classifier output, weighted and unweighted
# scores really should be unequal.
X, y = make_classification(random_state=0)
_, y_ml = make_multilabel_classification(n_samples=X.shape[0], random_state=0)
split = train_test_split(X, y, y_ml, random_state=0)
X_train, X_test, y_train, y_test, y_ml_train, y_ml_test = split
sample_weight = np.ones_like(y_test)
sample_weight[:10] = 0
# get sensible estimators for each metric
sensible_regr = DummyRegressor(strategy="median")
sensible_regr.fit(X_train, y_train)
sensible_clf = DecisionTreeClassifier(random_state=0)
sensible_clf.fit(X_train, y_train)
sensible_ml_clf = DecisionTreeClassifier(random_state=0)
sensible_ml_clf.fit(X_train, y_ml_train)
estimator = dict(
[(name, sensible_regr) for name in REGRESSION_SCORERS]
+ [(name, sensible_clf) for name in CLF_SCORERS]
+ [(name, sensible_ml_clf) for name in MULTILABEL_ONLY_SCORERS]
)
for name, scorer in SCORERS.items():
if name in MULTILABEL_ONLY_SCORERS:
target = y_ml_test
else:
target = y_test
try:
weighted = scorer(estimator[name], X_test, target, sample_weight=sample_weight)
ignored = scorer(estimator[name], X_test[10:], target[10:])
unweighted = scorer(estimator[name], X_test, target)
assert_not_equal(
weighted,
unweighted,
msg="scorer {0} behaves identically when "
"called with sample weights: {1} vs "
"{2}".format(name, weighted, unweighted),
)
assert_almost_equal(
weighted,
ignored,
err_msg="scorer {0} behaves differently when "
"ignoring samples and setting sample_weight to"
" 0: {1} vs {2}".format(name, weighted, ignored),
)
except TypeError as e:
assert_true(
"sample_weight" in str(e),
"scorer {0} raises unhelpful exception when called " "with sample weights: {1}".format(name, str(e)),
)
def test_median_strategy_regressor():
random_state = np.random.RandomState(seed=1)
X = [[0]] * 5 # ignored
y = random_state.randn(5)
reg = DummyRegressor(strategy="median")
reg.fit(X, y)
assert_array_equal(reg.predict(X), [np.median(y)] * len(X))
def test_dummy_regressor_return_std():
X = [[0]] * 3 # ignored
y = np.array([2, 2, 2])
y_std_expected = np.array([0, 0, 0])
cls = DummyRegressor()
cls.fit(X, y)
y_pred_list = cls.predict(X, return_std=True)
# there should be two elements when return_std is True
assert_equal(len(y_pred_list), 2)
# the second element should be all zeros
assert_array_equal(y_pred_list[1], y_std_expected)
def simplest(cube, y, cv):
""" just use the mean to impute the missing values
"""
from sklearn.dummy import DummyRegressor
clf = DummyRegressor()
X = cube.reshape(cube.shape[0], cube.shape[1] * cube.shape[2])
sse = np.zeros(y.shape[1])
for train, test in cv:
y_train, y_test = y[train], y[test]
y_predict = clf.fit(X[train], y[train]).predict(X[test])
sse += np.mean((y_predict - y_test) ** 2, 0)
return sse
def _make_estimators(X_train, y_train, y_ml_train):
# Make estimators that make sense to test various scoring methods
sensible_regr = DummyRegressor(strategy='median')
sensible_regr.fit(X_train, y_train)
sensible_clf = DecisionTreeClassifier(random_state=0)
sensible_clf.fit(X_train, y_train)
sensible_ml_clf = DecisionTreeClassifier(random_state=0)
sensible_ml_clf.fit(X_train, y_ml_train)
return dict(
[(name, sensible_regr) for name in REGRESSION_SCORERS] +
[(name, sensible_clf) for name in CLF_SCORERS] +
[(name, sensible_ml_clf) for name in MULTILABEL_ONLY_SCORERS]
)
def test_multioutput_regressor():
X_learn = np.random.randn(10, 10)
y_learn = np.random.randn(10, 5)
mean = np.mean(y_learn, axis=0).reshape((1, -1))
X_test = np.random.randn(20, 10)
y_test = np.random.randn(20, 5)
# Correctness oracle
est = DummyRegressor()
est.fit(X_learn, y_learn)
y_pred_learn = est.predict(X_learn)
y_pred_test = est.predict(X_test)
assert_array_equal(np.tile(mean, (y_learn.shape[0], 1)), y_pred_learn)
assert_array_equal(np.tile(mean, (y_test.shape[0], 1)), y_pred_test)
_check_behavior_2d(est)
def test_mean_strategy_multioutput_regressor():
random_state = np.random.RandomState(seed=1)
X_learn = random_state.randn(10, 10)
y_learn = random_state.randn(10, 5)
mean = np.mean(y_learn, axis=0).reshape((1, -1))
X_test = random_state.randn(20, 10)
y_test = random_state.randn(20, 5)
# Correctness oracle
est = DummyRegressor()
est.fit(X_learn, y_learn)
y_pred_learn = est.predict(X_learn)
y_pred_test = est.predict(X_test)
_check_equality_regressor(mean, y_learn, y_pred_learn, y_test, y_pred_test)
_check_behavior_2d(est)
def test_regressor_prediction_independent_of_X(strategy):
y = [0, 2, 1, 1]
X1 = [[0]] * 4
reg1 = DummyRegressor(strategy=strategy, constant=0, quantile=0.7)
reg1.fit(X1, y)
predictions1 = reg1.predict(X1)
X2 = [[1]] * 4
reg2 = DummyRegressor(strategy=strategy, constant=0, quantile=0.7)
reg2.fit(X2, y)
predictions2 = reg2.predict(X2)
assert_array_equal(predictions1, predictions2)
def test_scorer_sample_weight():
"""Test that scorers support sample_weight or raise sensible errors"""
# Unlike the metrics invariance test, in the scorer case it's harder
# to ensure that, on the classifier output, weighted and unweighted
# scores really should be unequal.
X, y = make_classification(random_state=0)
X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=0)
sample_weight = np.ones_like(y_test)
sample_weight[:10] = 0
# get sensible estimators for each metric
sensible_regr = DummyRegressor(strategy='median')
sensible_regr.fit(X_train, y_train)
sensible_clf = DecisionTreeClassifier()
sensible_clf.fit(X_train, y_train)
estimator = dict([(name, sensible_regr)
for name in REGRESSION_SCORERS] +
[(name, sensible_clf)
for name in CLF_SCORERS])
for name, scorer in SCORERS.items():
try:
weighted = scorer(estimator[name], X_test, y_test,
sample_weight=sample_weight)
ignored = scorer(estimator[name], X_test[10:], y_test[10:])
unweighted = scorer(estimator[name], X_test, y_test)
assert_not_equal(weighted, unweighted,
"scorer {0} behaves identically when called with "
"sample weights: {1} vs {2}".format(name,
weighted,
unweighted))
assert_equal(weighted, ignored,
"scorer {0} behaves differently when ignoring "
"samples and setting sample_weight to 0: "
"{1} vs {2}".format(name, weighted, ignored))
except TypeError as e:
assert_true("sample_weight" in str(e),
"scorer {0} raises unhelpful exception when called "
"with sample weights: {1}".format(name, str(e)))
def _minimize_simbo_general(fun,
x0, # only used to get number of features
args=(),
callback=None,
batch_size=100,
population_size=10000,
maxiter=10000,
scorer=None, # if no scorer given, scores are constant
selector=None, # only relevant is sampler is given
sampler=None):
n_iter = int(maxiter / batch_size)
assert n_iter > 0
dummy_generator = generative_models.DummyGenerator(len(x0))
if scorer is None:
scorer = DummyRegressor()
if sampler is None:
sampler = dummy_generator
if isinstance(selector, float) and 0 < selector < 1:
selector = percentile_selector(selector)
for i in range(n_iter):
if i == 0:
batch = dummy_generator.sample(batch_size)
else:
population = sampler.sample(population_size)
scores = scorer.predict(population)
batch_w_score = heapq.nsmallest(batch_size, zip(scores, population),
key=lambda x: x[0])
batch = [v for score, v in batch_w_score]
results = optimize_utils.score_multi(fun, batch, args, callback)
selected = selector(results, batch) if selector is not None else batch
scorer.fit(batch, results)
sampler.fit(selected)
best_fval, best_x = max(zip(results, batch), key=lambda x: x[0])
nfev = batch_size * n_iter
return optimize_utils.to_result(x=best_x, fun=best_fval,
niter=n_iter, nfev=nfev)
def test_constant_strategy_multioutput_regressor():
random_state = np.random.RandomState(seed=1)
X_learn = random_state.randn(10, 10)
y_learn = random_state.randn(10, 5)
# test with 2d array
constants = random_state.randn(5)
X_test = random_state.randn(20, 10)
y_test = random_state.randn(20, 5)
# Correctness oracle
est = DummyRegressor(strategy="constant", constant=constants)
est.fit(X_learn, y_learn)
y_pred_learn = est.predict(X_learn)
y_pred_test = est.predict(X_test)
_check_equality_regressor(constants, y_learn, y_pred_learn, y_test, y_pred_test)
_check_behavior_2d_for_constant(est)
def test_weights_regressor():
"""Check weighted average regression prediction on boston dataset."""
reg1 = DummyRegressor(strategy='mean')
reg2 = DummyRegressor(strategy='median')
reg3 = DummyRegressor(strategy='quantile', quantile=.2)
ereg = VotingRegressor([('mean', reg1), ('median', reg2),
('quantile', reg3)], weights=[1, 2, 10])
X_r_train, X_r_test, y_r_train, y_r_test = \
train_test_split(X_r, y_r, test_size=.25)
reg1_pred = reg1.fit(X_r_train, y_r_train).predict(X_r_test)
reg2_pred = reg2.fit(X_r_train, y_r_train).predict(X_r_test)
reg3_pred = reg3.fit(X_r_train, y_r_train).predict(X_r_test)
ereg_pred = ereg.fit(X_r_train, y_r_train).predict(X_r_test)
avg = np.average(np.asarray([reg1_pred, reg2_pred, reg3_pred]), axis=0,
weights=[1, 2, 10])
assert_almost_equal(ereg_pred, avg, decimal=2)
ereg_weights_none = VotingRegressor([('mean', reg1), ('median', reg2),
('quantile', reg3)], weights=None)
ereg_weights_equal = VotingRegressor([('mean', reg1), ('median', reg2),
('quantile', reg3)],
weights=[1, 1, 1])
ereg_weights_none.fit(X_r_train, y_r_train)
ereg_weights_equal.fit(X_r_train, y_r_train)
ereg_none_pred = ereg_weights_none.predict(X_r_test)
ereg_equal_pred = ereg_weights_equal.predict(X_r_test)
assert_almost_equal(ereg_none_pred, ereg_equal_pred, decimal=2)
def test_constant_strategy_regressor():
random_state = np.random.RandomState(seed=1)
X = [[0]] * 5 # ignored
y = random_state.randn(5)
reg = DummyRegressor(strategy="constant", constant=[43])
reg.fit(X, y)
assert_array_equal(reg.predict(X), [43] * len(X))
reg = DummyRegressor(strategy="constant", constant=43)
reg.fit(X, y)
assert_array_equal(reg.predict(X), [43] * len(X))
def test_stacked_featurizer(self):
data = self.make_test_data()
data['y'] = [1, 2, 3]
# Test for a regressor
model = DummyRegressor()
model.fit(self.multi.featurize_many(data['x']), data['y'])
# Test the predictions
f = StackedFeaturizer(self.single, model)
self.assertEqual([2], f.featurize(data['x'][0]))
# Test the feature names
self.assertEqual(['prediction'], f.feature_labels())
f.name = 'ML'
self.assertEqual(['ML prediction'], f.feature_labels())
# Test classifier
model = DummyClassifier("prior")
data['y'] = [0, 0, 1]
model.fit(self.multi.featurize_many(data['x']), data['y'])
# Test the prediction
f.model = model
self.assertEqual([2. / 3], f.featurize(data['x'][0]))
# Test the feature labels
self.assertRaises(ValueError, f.feature_labels)
f.class_names = ['A', 'B']
self.assertEqual(['ML P(A)'], f.feature_labels())
# Test with three classes
data['y'] = [0, 2, 1]
model.fit(self.multi.featurize_many(data['x']), data['y'])
self.assertArrayAlmostEqual([1. / 3] * 2, f.featurize(data['x'][0]))
f.class_names = ['A', 'B', 'C']
self.assertEqual(['ML P(A)', 'ML P(B)'], f.feature_labels())
def test_constant_strategy_multioutput_regressor():
random_state = np.random.RandomState(seed=1)
X_learn = random_state.randn(10, 10)
y_learn = random_state.randn(10, 5)
# test with 2d array
constants = random_state.randn(5)
X_test = random_state.randn(20, 10)
y_test = random_state.randn(20, 5)
# Correctness oracle
est = DummyRegressor(strategy="constant", constant=constants)
est.fit(X_learn, y_learn)
y_pred_learn = est.predict(X_learn)
y_pred_test = est.predict(X_test)
_check_equality_regressor(
constants, y_learn, y_pred_learn, y_test, y_pred_test)
_check_behavior_2d_for_constant(est)
def test_quantile_invalid():
X = [[0]] * 5 # ignored
y = [0] * 5 # ignored
est = DummyRegressor(strategy="quantile")
assert_raises(ValueError, est.fit, X, y)
est = DummyRegressor(strategy="quantile", quantile=None)
assert_raises(ValueError, est.fit, X, y)
est = DummyRegressor(strategy="quantile", quantile=[0])
assert_raises(ValueError, est.fit, X, y)
est = DummyRegressor(strategy="quantile", quantile=-0.1)
assert_raises(ValueError, est.fit, X, y)
est = DummyRegressor(strategy="quantile", quantile=1.1)
assert_raises(ValueError, est.fit, X, y)
est = DummyRegressor(strategy="quantile", quantile='abc')
assert_raises(TypeError, est.fit, X, y)
yelp['hate'] = yelp.text.str.contains('hate', case=False).astype(int)
# add new features to the model
feature_cols = ['cool', 'useful', 'funny', 'length', 'love', 'hate']
X = yelp[feature_cols]
train_test_rmse(X, y)
# TASK 8 (BONUS): compare your best RMSE with RMSE for the null model
# split the data (outside of the function)
X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=1)
# use scikit-learn's built-in dummy regressor
from sklearn.dummy import DummyRegressor
dumb = DummyRegressor(strategy='mean')
dumb.fit(X_train, y_train)
y_dumb = dumb.predict(X_test)
print np.sqrt(metrics.mean_squared_error(y_test, y_dumb))
# or, create a NumPy array with the right length, and fill it with the mean of y_train
y_null = np.zeros_like(y_test, dtype=float)
y_null.fill(y_train.mean())
print np.sqrt(metrics.mean_squared_error(y_test, y_null))
# TASK 9 (BONUS): treat this as a classification problem, try KNN, maximize your accuracy
from sklearn.neighbors import KNeighborsClassifier
knn = KNeighborsClassifier(n_neighbors=150)
knn.fit(X_train, y_train)
#data including the lines
test_season = pd.read_csv('data/test_season.csv')
train_season = pd.read_csv('data/train_season.csv')
X_train_s = train_season.drop('GAME_TOTAL', axis = 1).to_numpy()
y_train_s = train_season['GAME_TOTAL'].to_numpy()
X_test_s = test_season.drop('GAME_TOTAL', axis = 1).to_numpy()
y_test_s = test_season['GAME_TOTAL'].to_numpy()
Test_Vegas = test_season['TOTAL_CLOSE'].to_numpy()
Train_Vegas = train_season['TOTAL_CLOSE'].to_numpy()
#Vegas BASELINE = 17.650007402704748
mean_squared_error(np.append(y_train_s,y_test_s), np.append(Train_Vegas,Test_Vegas), squared = False)
#DUMMY REGRESSOR:
dummy_regr = DummyRegressor(strategy="mean")
dummy_regr.fit(X_train_s, y_train_s)
#-0.7833193001644205
dummy_regr.score(X_test_s, y_test_s)
#27.845427872989156
mean_squared_error(y_test_s, dummy_regr.predict(X_test_s), squared = False)
#OLS
regressor = sm.OLS(y_train_s, X_train_s)
regressor = regressor.fit()
#evidently this returned a 0.991 R**2
#second run gave us 0.993
regressor.summary()
preds = regressor.predict(X_test_s)
#18.5802074596655
mean_squared_error(y_test_s, preds, squared = False)
from sklearn.metrics import r2_score
from sklearn.metrics import mean_squared_error
import joblib
import sys
from pathlib import Path
sys.path.append('/home/jiajunb/prosocial-conversations')
from models import XGBOOST_FEATURES, EIGENMETRICS
ROOT_DIR = Path('/shared/0/projects/prosocial/data/finalized/')
train_df = pd.read_csv(ROOT_DIR / 'data_cache/lr_or_xgboost/train.tsv',
sep='\t',
usecols=XGBOOST_FEATURES + EIGENMETRICS)
train_X = train_df[XGBOOST_FEATURES].values
train_y = train_df[EIGENMETRICS].values.reshape(-1)
dummy_clf = DummyRegressor(strategy="mean")
dummy_clf.fit(train_X, train_y)
# on training set
train_preds = dummy_clf.predict(train_X)
print(f'R^2 on training set: {r2_score(train_y, train_preds)}')
print(f'MSELoss on training set: {mean_squared_error(train_preds, train_y)}')
output_path = ROOT_DIR / 'model_checkpoints/dummy'
output_path.mkdir(exist_ok=True, parents=True)
joblib.dump(dummy_clf, output_path / 'dummy.model.buffer')
test_df = pd.read_csv(ROOT_DIR / 'data_cache/lr_or_xgboost/test.tsv',
sep='\t',
usecols=XGBOOST_FEATURES + EIGENMETRICS)
test_X = test_df[XGBOOST_FEATURES].values
def DummyPrediction(X_train, y_train, X_test, y_test):
dummy = DummyRegressor()
dummy = dummy.fit(X_train, y_train)
y_pred = dummy.predict(X_test)
return (y_pred)
def test_regressor_exceptions():
reg = DummyRegressor()
assert_raises(ValueError, reg.predict, [])
def test_set_params_nested_pipeline():
estimator = Pipeline([('a', Pipeline([('b', DummyRegressor())]))])
estimator.set_params(a__b__alpha=0.001, a__b=Lasso())
estimator.set_params(a__steps=[('b', LogisticRegression())], a__b__C=5)
def main():
# load training and testing data set
print('parsing training set...')
X_train, y_train = parse('./data_set/train_set.csv')
print('parsing testing set...')
X_test, y_test = parse('./data_set/test_set.csv')
print('train set: ', X_train.shape)
print('test set: ', X_test.shape)
# The result turns out to be worse using non-linear polynomial regression
# convert to polynomial features
# print('converting to polynomial features...')
# poly = PolynomialFeatures(2)
# X_train = poly.fit_transform(X_train)
# X_test = poly.fit_transform(X_test)
# print('train set: ', X_train.shape)
# print('test set: ', X_test.shape)
# scale the attributes to [0, 1]
print('standardizing the features...')
min_max_scaler = MinMaxScaler()
X_train = min_max_scaler.fit_transform(X_train)
X_test = min_max_scaler.transform(X_test)
# training classifiers
print('training, predicting and evaluating...')
# Dummy Regression (baseline model)
print('\nDummy Regression: (baseline)')
model = DummyRegressor(strategy='mean')
model.fit(X_train, y_train)
y_pre = model.predict(X_test)
print('mean absolute error: ', mean_absolute_error(y_test, y_pre))
print('r2_score: ', r2_score(y_test, y_pre))
# Linear Regression
print('\nLinear_regression: ')
model = LinearRegression()
model.fit(X_train, y_train)
y_pre = model.predict(X_test)
print('mean absolute error: ', mean_absolute_error(y_test, y_pre))
print('r2_score: ', r2_score(y_test, y_pre))
# KNN Regression
# print('\nKNN Regression: ')
# model = KNeighborsRegressor()
# model.fit(X_train, y_train)
# y_pre = model.predict(X_test)
# print('mean absolute error: ', mean_absolute_error(y_test, y_pre))
# print('r2_score: ', r2_score(y_test, y_pre))
# Neural Network - Bernoulli Restricted Boltzmann Machine (RBM)
# print('\nNeural Network - RBM: ')
# model = BernoulliRBM()
# model.fit(X_train, y_train)
# y_pre = model.predict(X_test)
# print('mean absolute error: ', mean_absolute_error(y_test, y_pre))
# print('r2_score: ', r2_score(y_test, y_pre))
# AdaBoost
print('\nAdaBoost: ')
model = AdaBoostRegressor()
model.fit(X_train, y_train)
y_pre = model.predict(X_test)
print('mean absolute error: ', mean_absolute_error(y_test, y_pre))
print('r2_score: ', r2_score(y_test, y_pre))
# Random Forest
print('\nRandom Forest:')
model = RandomForestRegressor()
model.fit(X_train, y_train)
y_pre = model.predict(X_test)
print('mean absolute error: ', mean_absolute_error(y_test, y_pre))
print('r2_score: ', r2_score(y_test, y_pre))
def test_unknown_strategey_regressor():
X = [[0]] * 5
y = [1, 2, 4, 6, 8]
est = DummyRegressor(strategy='gona')
assert_raises(ValueError, est.fit, X, y)
def main():
# load training and testing data set
print('parsing training set...')
X_train, y_train = parse('./data_set/train_set.csv')
print('parsing testing set...')
X_test, y_test = parse('./data_set/test_set.csv')
print('train set: ', X_train.shape)
print('test set: ', X_test.shape)
# The result turns out to be worse using non-linear polynomial regression
# convert to polynomial features
# print('converting to polynomial features...')
# poly = PolynomialFeatures(2)
# X_train = poly.fit_transform(X_train)
# X_test = poly.fit_transform(X_test)
# print('train set: ', X_train.shape)
# print('test set: ', X_test.shape)
# scale the attributes to [0, 1]
print('standardizing the features...')
min_max_scaler = MinMaxScaler()
X_train = min_max_scaler.fit_transform(X_train)
X_test = min_max_scaler.transform(X_test)
# training classifiers
print('training, predicting and evaluating...')
# Dummy Regression (baseline model)
print('\nDummy Regression: (baseline)')
model = DummyRegressor(strategy='mean')
model.fit(X_train, y_train)
y_pre = model.predict(X_test)
print('mean absolute error: ', mean_absolute_error(y_test, y_pre))
print('r2_score: ', r2_score(y_test, y_pre))
# Linear Regression
print('\nLinear_regression: ')
model = LinearRegression()
model.fit(X_train, y_train)
y_pre = model.predict(X_test)
print('mean absolute error: ', mean_absolute_error(y_test, y_pre))
print('r2_score: ', r2_score(y_test, y_pre))
# KNN Regression
# print('\nKNN Regression: ')
# model = KNeighborsRegressor()
# model.fit(X_train, y_train)
# y_pre = model.predict(X_test)
# print('mean absolute error: ', mean_absolute_error(y_test, y_pre))
# print('r2_score: ', r2_score(y_test, y_pre))
# Neural Network - Bernoulli Restricted Boltzmann Machine (RBM)
# print('\nNeural Network - RBM: ')
# model = BernoulliRBM()
# model.fit(X_train, y_train)
# y_pre = model.predict(X_test)
# print('mean absolute error: ', mean_absolute_error(y_test, y_pre))
# print('r2_score: ', r2_score(y_test, y_pre))
# AdaBoost
print('\nAdaBoost: ')
model = AdaBoostRegressor()
model.fit(X_train, y_train)
y_pre = model.predict(X_test)
print('mean absolute error: ', mean_absolute_error(y_test, y_pre))
print('r2_score: ', r2_score(y_test, y_pre))
# Random Forest
print('\nRandom Forest:')
model = RandomForestRegressor()
model.fit(X_train, y_train)
y_pre = model.predict(X_test)
print('mean absolute error: ', mean_absolute_error(y_test, y_pre))
print('r2_score: ', r2_score(y_test, y_pre))
reg_drop = StackingRegressor(estimators=estimators,
final_estimator=rf,
cv=5)
reg.fit(X_train, y_train)
reg_drop.fit(X_train, y_train)
assert_allclose(reg.predict(X_test), reg_drop.predict(X_test))
assert_allclose(reg.transform(X_test), reg_drop.transform(X_test))
@pytest.mark.parametrize(
"cv", [3, KFold(n_splits=3, shuffle=True, random_state=42)])
@pytest.mark.parametrize("final_estimator, predict_params",
[(None, {}),
(RandomForestRegressor(random_state=42), {}),
(DummyRegressor(), {
'return_std': True
})])
@pytest.mark.parametrize("passthrough", [False, True])
def test_stacking_regressor_diabetes(cv, final_estimator, predict_params,
passthrough):
# prescale the data to avoid convergence warning without using a pipeline
# for later assert
X_train, X_test, y_train, _ = train_test_split(scale(X_diabetes),
y_diabetes,
random_state=42)
estimators = [('lr', LinearRegression()), ('svr', LinearSVR())]
reg = StackingRegressor(estimators=estimators,
final_estimator=final_estimator,
cv=cv,
passthrough=passthrough)
from sklearn.metrics import mean_squared_error
# In[35]:
X_train, X_test, y_train, y_test = train_test_split(X,
y,
test_size=0.2,
random_state=42)
# # Baseline Model
# In[36]:
from sklearn.dummy import DummyRegressor
dummy_regr = DummyRegressor(strategy="mean")
dummy_regr.fit(X_train, y_train)
dummy_regr.predict(X_train)
baseline = dummy_regr.score(X_train, y_train)
print("Baseline R^2: %f" % baseline)
# # Multiple Linear Regression
# In[37]:
ols = linear_model.LinearRegression()
ols.fit(X_train, y_train)
print("Coefficients: %s" % ols.coef_)
print("Intercept: %f" % ols.intercept_)
y_test_prediction = ols.predict(X_test)
ols.score(X_train, y_train)
def test_regressor_score_with_None(y, y_test):
reg = DummyRegressor()
reg.fit(None, y)
assert reg.score(None, y_test) == 1.0
import numpy as np
from sklearn.model_selection import train_test_split
from sklearn import datasets
from sklearn.linear_model import LinearRegression
from sklearn.metrics import mean_squared_error, r2_score
from sklearn.dummy import DummyRegressor
diabetes = datasets.load_diabetes()
X = diabetes.data[:, None, 6]
y = diabetes.target
X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=0)
lm = LinearRegression().fit(X_train, y_train)
lm_dummy_mean = DummyRegressor(strategy = 'mean').fit(X_train, y_train)
y_predict = lm.predict(X_test)
y_predict_dummy_mean = lm_dummy_mean.predict(X_test)
print('Linear model, coefficients: ', lm.coef_)
print("Mean squared error (dummy): {:.2f}".format(mean_squared_error(y_test,
y_predict_dummy_mean)))
print("Mean squared error (linear model): {:.2f}".format(mean_squared_error(y_test, y_predict)))
print("r2_score (dummy): {:.2f}".format(r2_score(y_test, y_predict_dummy_mean)))
print("r2_score (linear model): {:.2f}".format(r2_score(y_test, y_predict)))
# Plot outputs
plt.scatter(X_test, y_test, color='black')
plt.plot(X_test, y_predict, color='green', linewidth=2)
plt.plot(X_test, y_predict_dummy_mean, color='red', linestyle = 'dashed',
import csv
import pickle
from sklearn.dummy import DummyRegressor
import numpy as np
age_range = 80
gender = {'male': 0, 'other': 0.5, 'female': 1}
X = np.array([[20 / age_range, gender['male']], [56 / age_range, gender['other']]])
Y = np.array([[.2] , [.7]])
clf = DummyRegressor()
clf.fit(X, Y)
# print([r[2] for r in data])
print(Y)
# print([
# movies[int(round(idx * len(movies)))]
# for idx in clf.predict(X)
# ])
print(clf.predict([[0.2, 1]]))
with open('model.pk', 'wb') as outfile:
pickle.dump(clf, outfile)
except ImportError:
# for scikit-learn 0.18 and 0.19
from sklearn.metrics.scorer import check_scoring
# Regression
ridge = RidgeCV()
svr = SVR(kernel='linear')
# Classification
svc = LinearSVC()
logistic_l1 = LogisticRegression(penalty='l1')
logistic_l2 = LogisticRegression(penalty='l2')
ridge_classifier = RidgeClassifierCV()
random_forest = RandomForestClassifier()
dummy_classifier = DummyClassifier(random_state=0)
dummy_regressor = DummyRegressor()
regressors = {'ridge': (ridge, []),
'svr': (svr, 'C')}
classifiers = {'svc': (svc, 'C'),
'logistic_l1': (logistic_l1, 'C'),
'logistic_l2': (logistic_l2, 'C'),
'ridge_classifier': (ridge_classifier, [])}
# Create a test dataset
rng = np.random.RandomState(0)
X = rng.rand(100, 10)
# Create different targets
y_regression = rng.rand(100)
y_classification = np.hstack([[-1] * 50, [1] * 50])
y_classification_str = np.hstack([['face'] * 50, ['house'] * 50])
y_multiclass = np.hstack([[0] * 35, [1] * 30, [2] * 35])
# Wczytanie bibliotek.
from sklearn.datasets import load_boston
from sklearn.dummy import DummyRegressor
from sklearn.model_selection import train_test_split
# Wczytanie danych.
boston = load_boston()
# Utworzenie cech.
features, target = boston.data, boston.target
# Podział na zbiory uczący i testowy.
features_train, features_test, target_train, target_test = train_test_split(
features, target, random_state=0)
# Utworzenie sztucznego regresora.
dummy = DummyRegressor(strategy='mean')
# "Wytrenowanie" sztucznego regresora.
dummy.fit(features_train, target_train)
# Pobranie kwadratu wartości.
dummy.score(features_test, target_test)
def test_quantile_strategy_regressor():
random_state = np.random.RandomState(seed=1)
X = [[0]] * 5 # ignored
y = random_state.randn(5)
reg = DummyRegressor(strategy="quantile", quantile=0.5)
reg.fit(X, y)
assert_array_equal(reg.predict(X), [np.median(y)] * len(X))
reg = DummyRegressor(strategy="quantile", quantile=0)
reg.fit(X, y)
assert_array_equal(reg.predict(X), [np.min(y)] * len(X))
reg = DummyRegressor(strategy="quantile", quantile=1)
reg.fit(X, y)
assert_array_equal(reg.predict(X), [np.max(y)] * len(X))
reg = DummyRegressor(strategy="quantile", quantile=0.3)
reg.fit(X, y)
assert_array_equal(reg.predict(X), [np.percentile(y, q=30)] * len(X))
def base_dummy(self):
model = DummyRegressor(strategy='mean')
setattr(model, 'data_schema', self.data._X.columns.values)
setattr(model, 'model_path', 'model_path')
return model.fit(self.data._X, self.data._y)
def test_quantile_strategy_empty_train():
est = DummyRegressor(strategy="quantile", quantile=0.4)
assert_raises(ValueError, est.fit, [], [])
def __init__(self):
self.clf = DummyRegressor()
def test_regressor_score_with_None(y, y_test):
reg = DummyRegressor()
reg.fit(None, y)
assert_equal(reg.score(None, y_test), 1.0)
# be significantly imbalanced, and even a simplistic model that would only
# predict mean can achieve an accuracy of 93%.
#
# To evaluate the pertinence of the used metrics, we will consider as a
# baseline a "dummy" estimator that constantly predicts the mean frequency of
# the training sample.
from sklearn.dummy import DummyRegressor
from sklearn.pipeline import Pipeline
from sklearn.model_selection import train_test_split
df_train, df_test = train_test_split(df, test_size=0.33, random_state=0)
dummy = Pipeline([
("preprocessor", linear_model_preprocessor),
("regressor", DummyRegressor(strategy="mean")),
]).fit(df_train,
df_train["Frequency"],
regressor__sample_weight=df_train["Exposure"])
##############################################################################
# Let's compute the performance of this constant prediction baseline with 3
# different regression metrics:
from sklearn.metrics import mean_squared_error
from sklearn.metrics import mean_absolute_error
from sklearn.metrics import mean_poisson_deviance
def score_estimator(estimator, df_test):
"""Score an estimator on the test set."""
def test_init(self):
regressor = DummyRegressor()
regressor.fit(numpy.array([[0], [0]]), numpy.array([0.0, 2.0]))
self.assertEqual(1.0, regressor.constant_)
regressor_proxy = EstimatorProxy(regressor, attr_names_=["constant_"])
self.assertEqual(1.0, regressor_proxy.constant_)
def test_quantile_strategy_multioutput_regressor():
random_state = np.random.RandomState(seed=1)
X_learn = random_state.randn(10, 10)
y_learn = random_state.randn(10, 5)
median = np.median(y_learn, axis=0).reshape((1, -1))
quantile_values = np.percentile(y_learn, axis=0, q=80).reshape((1, -1))
X_test = random_state.randn(20, 10)
y_test = random_state.randn(20, 5)
# Correctness oracle
est = DummyRegressor(strategy="quantile", quantile=0.5)
est.fit(X_learn, y_learn)
y_pred_learn = est.predict(X_learn)
y_pred_test = est.predict(X_test)
_check_equality_regressor(
median, y_learn, y_pred_learn, y_test, y_pred_test)
_check_behavior_2d(est)
# Correctness oracle
est = DummyRegressor(strategy="quantile", quantile=0.8)
est.fit(X_learn, y_learn)
y_pred_learn = est.predict(X_learn)
y_pred_test = est.predict(X_test)
_check_equality_regressor(
quantile_values, y_learn, y_pred_learn, y_test, y_pred_test)
_check_behavior_2d(est)
def main():
# read review data
print('parsing review data...')
reviews = parse_json('./yelp_dataset_challenge_academic_dataset/yelp_academic_dataset_review.json')
# use only reviews posted after 2008
valid_reviews = []
for review in reviews:
review_date = datetime.datetime.strptime(review['date'], '%Y-%m-%d')
if review_date.year < 2008:
continue
valid_reviews.append(review)
reviews = valid_reviews
# sample the data
# sample_num = len(reviews)
# print('sampling...', sample_num, 'out of', len(reviews))
# reviews = sample(reviews, sample_num)
# tokenize text for all reviews
print('tokenizing text for all reviews...')
texts = [review['text'] for review in reviews]
count_vect = CountVectorizer(max_features = 100)
X = count_vect.fit_transform(texts)
# transform from occurrence to frequency
print('converting occurrence to frequency...')
tfidf_transformer = TfidfTransformer()
X = tfidf_transformer.fit_transform(X)
# load the linear model for normalization
clf = joblib.load('./normalization/linear_model_for_normalization.pkl')
# get labels
print('calculating labels...')
y = []
for review in reviews:
review_date = datetime.datetime.strptime(review['date'], '%Y-%m-%d')
# normalize
normalizor = clf.predict(np.array([[review_date.year]]))[0][0]
review_quality = sum(review['votes'].values()) / normalizor
y.append(review_quality)
# splitting into train and test set
print('splitting into train and test set...')
train_len = int(X.shape[0] * 0.6)
X_train = X[:train_len, :]
y_train = y[:train_len]
X_test = X[train_len:, :]
y_test = y[train_len:]
print('train size:', X_train.shape)
print('test size:', X_test.shape)
# convert to polynomial features
# print('converting to polynomial features...')
# poly = PolynomialFeatures(2)
# X_train = poly.fit_transform(X_train.toarray())
# X_test = poly.fit_transform(X_test.toarray())
# print('train set: ', X_train.shape)
# print('test set: ', X_test.shape)
# scale the attributes to [0, 1]
print('standardizing the features...')
min_max_scaler = MinMaxScaler()
X_train = min_max_scaler.fit_transform(X_train)
X_test = min_max_scaler.transform(X_test)
# training classifiers
print('training, predicting and evaluating...')
# Dummy Regression (baseline model)
print('\nDummy Regression:')
model = DummyRegressor(strategy='mean')
model.fit(X_train, y_train)
y_pre = model.predict(X_test)
print('mean absolute error: ', mean_absolute_error(y_test, y_pre))
print('r2_score: ', r2_score(y_test, y_pre))
# Linear Regression
print('\nLinear_regression: ')
model = LinearRegression()
model.fit(X_train, y_train)
y_pre = model.predict(X_test)
print('mean absolute error: ', mean_absolute_error(y_test, y_pre))
print('r2_score: ', r2_score(y_test, y_pre))
# Ridge
print('\nRidge: ')
model = Ridge()
model.fit(X_train, y_train)
y_pre = model.predict(X_test)
print('mean absolute error: ', mean_absolute_error(y_test, y_pre))
print('r2_score: ', r2_score(y_test, y_pre))
# passive aggresive
print('\nPoly: ')
model = PassiveAggressiveRegressor()
model.fit(X_train, y_train)
y_pre = model.predict(X_test)
print('mean absolute error: ', mean_absolute_error(y_test, y_pre))
print('r2_score: ', r2_score(y_test, y_pre))
# AdaBoost
print('\nAdaBoost: ')
model = AdaBoostRegressor()
model.fit(X_train, y_train)
y_pre = model.predict(X_test)
print('mean absolute error: ', mean_absolute_error(y_test, y_pre))
print('r2_score: ', r2_score(y_test, y_pre))
# Random Forest
print('\nRandom Forest:')
model = RandomForestRegressor()
model.fit(X_train, y_train)
y_pre = model.predict(X_test)
print('mean absolute error: ', mean_absolute_error(y_test, y_pre))
print('r2_score: ', r2_score(y_test, y_pre))
print("Percentage of zero claims = {0:%}".format(
df.loc[df["ClaimNb"] == 0, "Exposure"].sum() / df["Exposure"].sum()))
##############################################################################
# It is worth noting that 92 % of policyholders have zero claims, and if we
# were to convert this problem into a binary classification task, it would be
# significantly imbalanced.
#
# To evaluate the pertinence of the used metrics, we will consider as a
# baseline a "dummy" estimator that constantly predicts the mean frequency of
# the training sample.
df_train, df_test = train_test_split(df, random_state=0)
dummy = make_pipeline(linear_model_preprocessor,
DummyRegressor(strategy='mean'))
dummy.fit(df_train,
df_train["Frequency"],
dummyregressor__sample_weight=df_train["Exposure"])
def score_estimator(estimator, df_test):
"""Score an estimator on the test set."""
y_pred = estimator.predict(df_test)
print(
"MSE: %.3f" %
mean_squared_error(df_test["Frequency"], y_pred, df_test["Exposure"]))
print(
"MAE: %.3f" %
from sklearn.metrics import mean_absolute_error
from sklearn.dummy import DummyRegressor
nobs = X_meg.shape[0]
max_comps = range(2,30,2)
nfolds=50
cv = ShuffleSplit(nobs,n_iter=nfolds,test_size=.1)
# Trying the prediction with different components
comp_scores = []
dumb_scores = []
for ncomp in max_comps:
print 'Trying %d components'%ncomp
pls = PLSRegression(n_components=ncomp)
dumb = DummyRegressor(strategy='mean')
mae = 0
dumb_mae = 0
for oidx, (train, test) in enumerate(cv):
X_fmri_train = X_fmri[train]
X_fmri_test = X_fmri[test]
X_meg_train = X_meg[train]
X_meg_test = X_meg[test]
pls.fit(X_fmri_train, X_meg_train)
pred = pls.predict(X_fmri_test)
mae += mean_absolute_error(X_meg_test, pred)
dumb.fit(X_fmri_train, X_meg_train)
def test_constants_not_specified_regressor():
X = [[0]] * 5
y = [1, 2, 4, 6, 8]
est = DummyRegressor(strategy='constant')
assert_raises(TypeError, est.fit, X, y)
# prepare configuration for cross validation test harness
num_folds = 10
seed = 7
# prepare models
models = []
models.append(('LR', LinearRegression()))
models.append(('Ridge', Ridge()))
#models.append(('ARDRegression', linear_model.ARDRegression()))
models.append(('Lasso', linear_model.Lasso()))
models.append(('LassoCV', linear_model.LassoCV()))
models.append(('LassoLars', linear_model.LassoLars()))
# Decision tree
models.append(('Dec tree', tree.DecisionTreeRegressor()))
# sanity check
models.append(('Dummy', DummyRegressor("median")))
def keras_baseline_model():
# create model
model = Sequential()
model.add(
Dense(128, input_dim=numFeatures, init='normal', activation='relu'))
model.add(Dense(1, init='normal', activation="relu"))
# Compile model
model.compile(loss='mean_squared_error', optimizer='adam')
return model
models.append(('Keras',
KerasRegressor(build_fn=keras_baseline_model,
pipeline.configure(**pmml_options) if isinstance(regressor, XGBRegressor): pipeline.verify(auto_X.sample(frac = 0.05, random_state = 13), predict_params = predict_params, precision = 1e-5, zeroThreshold = 1e-5) else: pipeline.verify(auto_X.sample(frac = 0.05, random_state = 13), predict_params = predict_params) store_pkl(pipeline, name) mpg = DataFrame(pipeline.predict(auto_X, **predict_params), columns = ["mpg"]) store_csv(mpg, name) if "Auto" in datasets: build_auto(AdaBoostRegressor(DecisionTreeRegressor(min_samples_leaf = 5, random_state = 13), random_state = 13, n_estimators = 17), "AdaBoostAuto") build_auto(ARDRegression(normalize = True), "BayesianARDAuto") build_auto(BayesianRidge(normalize = True), "BayesianRidgeAuto") build_auto(DecisionTreeRegressor(min_samples_leaf = 2, random_state = 13), "DecisionTreeAuto", compact = False) build_auto(BaggingRegressor(DecisionTreeRegressor(min_samples_leaf = 5, random_state = 13), n_estimators = 3, max_features = 0.5, random_state = 13), "DecisionTreeEnsembleAuto") build_auto(DummyRegressor(strategy = "median"), "DummyAuto") build_auto(ElasticNetCV(cv = 3, random_state = 13), "ElasticNetAuto") build_auto(ExtraTreesRegressor(n_estimators = 10, min_samples_leaf = 5, random_state = 13), "ExtraTreesAuto") build_auto(GBDTLMRegressor(RandomForestRegressor(n_estimators = 7, max_depth = 6, random_state = 13), LinearRegression()), "GBDTLMAuto") build_auto(GBDTLMRegressor(XGBRFRegressor(n_estimators = 17, max_depth = 6, random_state = 13), ElasticNet(random_state = 13)), "XGBRFLMAuto") build_auto(GradientBoostingRegressor(init = None, random_state = 13), "GradientBoostingAuto") build_auto(HistGradientBoostingRegressor(max_iter = 31, random_state = 13), "HistGradientBoostingAuto") build_auto(HuberRegressor(), "HuberAuto") build_auto(LarsCV(cv = 3), "LarsAuto") build_auto(LassoCV(cv = 3, random_state = 13), "LassoAuto") build_auto(LassoLarsCV(cv = 3), "LassoLarsAuto") build_auto(LinearRegression(), "LinearRegressionAuto") build_auto(BaggingRegressor(LinearRegression(), max_features = 0.75, random_state = 13), "LinearRegressionEnsembleAuto") build_auto(OrthogonalMatchingPursuitCV(cv = 3), "OMPAuto") build_auto(RandomForestRegressor(n_estimators = 10, min_samples_leaf = 3, random_state = 13), "RandomForestAuto", flat = True) build_auto(RidgeCV(), "RidgeAuto")
import pandas as pd
from sklearn.dummy import DummyRegressor
# Loading in the data
canucks = pd.read_csv('data/canucks_subbed.csv')
# Define X and y
X = canucks.loc[:, ['No.', 'Age', 'Height', 'Weight', 'Experience']]
y = canucks['Salary']
# Create a model
model = DummyRegressor(strategy="mean")
# Fit your data
model.fit(X, y)
# Predict the labels of X
model.predict(X)
# The model accuracy
accuracy = round(model.score(X, y), 2)
accuracy
df = test_regressor(ExtraTreesRegressor(n_estimators=1000), df)
df = test_regressor(GradientBoostingRegressor(n_estimators=1000), df)
df = test_regressor(RandomForestRegressor(n_estimators=1000), df)
df = test_regressor(GaussianProcessRegressor(), df)
# df = test_regressor(IsotonicRegression(), df) - has errors
df = test_regressor(LinearSVR(), df)
df = test_regressor(NuSVR(), df)
df = test_regressor(SVR(), df)
df = test_regressor(XGBRegressor(n_estimators=1000), df)
df = test_regressor(lgb.LGBMRegressor(n_estimators=1000), df)
df = test_regressor(CatBoostRegressor(n_estimators=1000), df)
df = test_regressor(DecisionTreeRegressor(max_depth=3), df)
df = test_regressor(KNeighborsRegressor(), df)
# df = test_regressor(RadiusNeighborsRegressor(), df) - also has errors
df = test_regressor(DummyRegressor(), df)
df = test_regressor(
StackingRegressor(regressors=[
GradientBoostingRegressor(n_estimators=1000),
HuberRegressor(),
RidgeCV(cv=5),
BayesianRidge(compute_score=True, copy_X=True)
],
meta_regressor=LassoCV(cv=5)), df)
df = test_regressor(
StackingRegressor(regressors=[
ElasticNetCV(),
HuberRegressor(),
RidgeCV(cv=5),
from sklearn.dummy import DummyRegressor
nobs = X_meg.shape[0]
max_comps = range(5,30,5)
nfolds=50
cv = ShuffleSplit(nobs,n_iter=nfolds,test_size=.1)
y = inatt
# Trying the prediction with different components
comp_scores = []
dumb_scores = []
meg_scores, fmri_scores = [], []
for ncomp in max_comps:
print 'Trying %d components'%ncomp
pls = PLSRegression(n_components=ncomp)
dumb = DummyRegressor(strategy='mean')
mae = 0
dumb_mae = 0
meg_mae, fmri_mae = 0, 0
for oidx, (train, test) in enumerate(cv):
X_fmri_train = X_fmri[train]
X_fmri_test = X_fmri[test]
X_meg_train = X_meg[train]
X_meg_test = X_meg[test]
y_train = y[train]
y_test = y[test]
X_train = np.hstack([X_fmri_train,X_meg_train])
X_test = np.hstack([X_fmri_test,X_meg_test])
def fit(self, X, y):
self.reg = DummyRegressor()
return self.reg.fit(X, y)
# 'gb': GradientBoostingClassifier(),
'xgb': XGBClassifier(),
'dummy': DummyClassifier()
}
REGRESSORS = {
'lr': LinearRegression(),
'lasso': Lasso(),
'ridge': Ridge(),
'mlp': MLPRegressor(),
'SVC': svm.SVR(),
'knn': KNeighborsRegressor(),
'rf': RandomForestRegressor(),
'gb': GradientBoostingRegressor(),
'xgb': XGBRegressor(),
'dummy': DummyRegressor()
}
def spotcheck(estimators=CLASSIFIERS,
X=None,
y=None,
score='roc_auc',
cv=3,
sort_by='mean'):
logging.info("Evaluation metrics: " + str(score))
results = {}
@timeit(get_time=True)
def _eval(clf):
scores = cross_val_score(estimator=clf, X=X, y=y, scoring=score, cv=cv)
