def LassoICReg(df, z, Ncy0, Ty0):
reg = linear_model.LassoLarsIC(criterion='bic')
reg2 = linear_model.LassoLarsIC(criterion='bic')
X = df.loc[:, 'ya0':'Tx']
y = df.loc[:, 'Ty']
y2 = df.loc[:, 'Ncy']
#z= df.loc[:,'ya0':'Tx']
reg.fit(X, y)
reg2.fit(X, y2)
predicted = reg.predict(z)
predicted2 = reg2.predict(z)
print('-----------------------------')
print("Regressão de Lasso IC (BIC)")
print(print('Ty: ', predicted))
print(print('Ncy: ', predicted2))
index = []
for ind in range(0, len(z.index)):
index.append(ind)
d = {
'index': index,
'Ty': predicted,
'Ncy': predicted2,
'Ncy0': Ncy0,
'Ty0': Ty0
}
df2 = pd.DataFrame(data=d)
return df2
def test_no_warning_for_zero_mse():
# LassoLarsIC should not warn for log of zero MSE.
y = np.arange(10, dtype=float)
X = y.reshape(-1, 1)
lars = linear_model.LassoLarsIC(normalize=False)
assert_no_warnings(lars.fit, X, y)
assert_true(np.any(np.isinf(lars.criterion_)))
def selector(ts_name, lang, a_df, r_df, cutoff_date):
# for each language, include only regressors that correlate well with the time series to forecast
# use lasso to find out which ones are to be kept
r_list = [c for c in r_df.columns if c not in ['ds', 'language']]
ycol = [c for c in a_df.columns if c not in ['ds', 'language']][0]
if len(a_df) == 0:
s_ut.my_print('WARNING: empty actuals for ' + lang + ' and ts ' +
ts_name)
return None
if len(r_df) == 0:
s_ut.my_print('WARNING: empty regressors for ' + lang + ' and ts ' +
ts_name)
return None
ds_min = max(a_df['ds'].min(), r_df['ds'].min())
s_df = r_df[(r_df['ds'] >= ds_min) & (r_df['ds'] <= cutoff_date)].copy()
b_df = a_df[(a_df['ds'] >= ds_min) & (a_df['ds'] <= cutoff_date)].copy()
df = b_df.merge(s_df, on=['ds', 'language'], how='left')
df.dropna(inplace=True)
X_train = df[r_list].values
y_train = df[ycol].values
lasso_mdl = l_mdl.LassoLarsIC(criterion='aic', normalize=True)
lasso_mdl.fit(X_train, y_train)
m_pars = lasso_mdl.coef_ # pars excluding intercept
new_r_list = [r_list[i] for i in range(len(r_list)) if m_pars[i] != 0.0]
s_ut.my_print('regressor selector::regressors for ' + ts_name +
' and language ' + lang + ': ' + str(new_r_list))
return r_df[['ds', 'language'] +
new_r_list].copy() if len(new_r_list) > 0 else None
def lasso_selection(lang, lang_f, af, normalize=True):
cfg_list = lang_f['cfg_idx'].unique()
f = lang_f[lang_f['cfg_idx'].isin(cfg_list)]
if len(f) == 0:
s_ut.my_print('pid: ' + str(os.getpid()) + ' WARNING: no data for cfg_list: ' + str(cfg_list))
return None
pf = pd.pivot_table(f[['ds', 'yhat', 'cfg_idx', 'language']], index=['ds', 'language'], columns='cfg_idx', values='yhat').reset_index(level=0).reset_index(drop=True)
pf.dropna(inplace=True) # ensure all fcast cfgs have the same time range
df = pf.merge(af, on='ds', how='inner')
pars = len(cfg_list) # regression parameters: one per fcast_cfg plus 3 extra pars: constant + 2 regularization
ndata = len(df) # data points
while ndata < pars: # more regression unknowns than data
s_ut.my_print('pid: ' + str(os.getpid()) + ' WARNING: not enough data for cfg_list: ' + str(cfg_list))
cfg_list = cfg_list[:int(0.75 * ndata)] # pick the top one by score
pars = len(cfg_list)
if len(cfg_list) == 0:
s_ut.my_print('pid: ' + str(os.getpid()) + ' ERROR: not enough data for cfg_list: ' + str(cfg_list))
return None
X_train = df[cfg_list].values
y_train = df['y'].values
lasso_mdl = l_mdl.LassoLarsIC(criterion='aic', normalize=normalize)
lasso_mdl.fit(X_train, y_train)
m_pars = lasso_mdl.coef_ # pars excluding intercept
new_cfg_list = [int(cfg_list[i]) for i in range(len(cfg_list)) if m_pars[i] != 0.0]
return {'cfg_idx': new_cfg_list, 'alpha': lasso_mdl.alpha_, 'language': lang, 'normalize': normalize}
def _regression_B(self, X):
residual_flag_ = []
causal_matrix = self.B_prune.copy()
reg_list = {i: causal_matrix[i, :] != 0 for i in range(self.n_dim)}
for i in range(self.n_dim):
if np.sum(reg_list[i]) != 0:
y_reg = X[:, i]
X_reg = X.T[reg_list[i]].T
if self.reg_type == "lasso":
clf = linear_model.LassoLarsIC(criterion=self.criterion,
max_iter=self.max_iter_ls,
precompute="auto")
else:
clf = linear_model.LinearRegression()
clf.fit(y=y_reg.reshape(self.n_samples, -1),
X=X_reg.reshape(self.n_samples, -1))
residual = y_reg.reshape(self.n_samples, -1) - clf.predict(
X_reg.reshape(self.n_samples, -1))
if self.shapiro == True:
if stats.shapiro(residual)[1] > 0.05:
norm_flag = True
else:
norm_flag = False
residual_flag_.append(norm_flag)
causal_matrix[i, reg_list[i]] = clf.coef_
else:
norm_flag = False
residual_flag_.append(norm_flag)
self.residual_flag = residual_flag_
return causal_matrix
def choose_optimizer(self,
LassoType='Lasso',
RegCoef=0.00001,
cv=5,
criterion='aic',
maxiter=10000,
tolerance=0.0001,
normalize=True):
if LassoType == 'Lasso':
lin = linear_model.Lasso(alpha=RegCoef,
max_iter=maxiter,
normalize=normalize,
tol=tolerance)
elif LassoType == 'LassoCV':
lin = linear_model.LassoCV(cv=cv,
normalize=normalize,
max_iter=maxiter)
elif LassoType == 'LassoLarsCV':
lin = linear_model.LassoLarsCV(cv=cv,
normalize=normalize,
max_iter=maxiter)
elif LassoType == 'LarsCV':
lin = linear_model.LarsCV(cv=cv,
normalize=normalize,
max_iter=maxiter)
elif LassoType == 'LassoLarsIC':
lin = linear_model.LassoLarsIC(criterion=criterion,
normalize=normalize,
max_iter=maxiter)
else:
raise Exception("wrong option")
return lin
def test_sk_LassoLarsIC():
print("Testing sklearn, LassoLarsIC...")
mod = linear_model.LassoLarsIC()
X, y = iris_data
mod.fit(X, y)
docs = {'name': "LassoLarsIC test"}
fv = X[0, :]
upload(mod, fv, docs)
def test_lasso_lars_ic():
""" Test the LassoLarsIC object by checking that
- some good features are selected.
- alpha_bic > alpha_aic
- n_nonzero_bic < n_nonzero_aic
"""
lars_bic = linear_model.LassoLarsIC('bic')
lars_aic = linear_model.LassoLarsIC('aic')
rng = np.random.RandomState(42)
X = diabetes.data
y = diabetes.target
X = np.c_[X, rng.randn(X.shape[0], 4)] # add 4 bad features
lars_bic.fit(X, y)
lars_aic.fit(X, y)
nonzero_bic = np.where(lars_bic.coef_)[0]
nonzero_aic = np.where(lars_aic.coef_)[0]
assert_greater(lars_bic.alpha_, lars_aic.alpha_)
assert_less(len(nonzero_bic), len(nonzero_aic))
assert_less(np.max(nonzero_bic), diabetes.data.shape[1])
def bootstrap(miR_exp, mRNA_exp, num_mrna):
clf = linear_model.LassoLarsIC(criterion='bic')
boot_coll = []
std = []
for i in range(1000):
rand_mrna = np.transpose(np.random.permutation(np.transpose(mRNA_exp)))
clf.fit(np.transpose(rand_mrna), np.transpose(miR_exp))
boot_coll.append(clf.coef_)
stdev = np.std(boot_coll, axis=0)
return stdev
def test_model_lasso_lars_ic(self):
model, X = fit_regression_model(linear_model.LassoLarsIC())
model_onnx = convert_sklearn(
model,
"lasso lars cv", [("input", FloatTensorType([None, X.shape[1]]))],
target_opset=TARGET_OPSET)
self.assertIsNotNone(model_onnx)
dump_data_and_model(X,
model,
model_onnx,
basename="SklearnLassoLarsIC-Dec4")
def test_lasso_lars_ic():
# Test the LassoLarsIC object by checking that
# - some good features are selected.
# - alpha_bic > alpha_aic
# - n_nonzero_bic < n_nonzero_aic
lars_bic = linear_model.LassoLarsIC('bic')
lars_aic = linear_model.LassoLarsIC('aic')
rng = np.random.RandomState(42)
X = diabetes.data
X = np.c_[X, rng.randn(X.shape[0], 5)] # add 5 bad features
lars_bic.fit(X, y)
lars_aic.fit(X, y)
nonzero_bic = np.where(lars_bic.coef_)[0]
nonzero_aic = np.where(lars_aic.coef_)[0]
assert lars_bic.alpha_ > lars_aic.alpha_
assert len(nonzero_bic) < len(nonzero_aic)
assert np.max(nonzero_bic) < diabetes.data.shape[1]
# test error on unknown IC
lars_broken = linear_model.LassoLarsIC('<unknown>')
assert_raises(ValueError, lars_broken.fit, X, y)
def regressionMethods(independent, dependent, regType=0): if regType == 0: clf = linear_model.RidgeCV(alphas=[0.1, 1.0, 10.0]) elif regType == 1: clf = linear_model.LassoCV(alphas=[0.1, 1.0, 10.0]) elif regtype == 2: clf = linear_model.LassoLarsIC(criterion='bic') elif regType == 3: clf = linear_model.ElasticNetCV(alphas=[0.1, 1.0, 10.0]) clf.fit (independent, dependent) return clf
def test_model_lasso_lars_ic(self):
model, X = _fit_model(linear_model.LassoLarsIC())
model_onnx = convert_sklearn(model, "lasso lars cv",
[("input", FloatTensorType(X.shape))])
self.assertIsNotNone(model_onnx)
dump_data_and_model(
X.astype(numpy.float32),
model,
model_onnx,
basename="SklearnLassoLarsIC-Dec4",
allow_failure="StrictVersion("
"onnxruntime.__version__)"
"<= StrictVersion('0.2.1')",
)
def test_model_lasso_lars_ic(self):
model, X = fit_regression_model(linear_model.LassoLarsIC())
model_onnx = convert_sklearn(
model,
"lasso lars cv", [("input", FloatTensorType([None, X.shape[1]]))],
target_opset=TARGET_OPSET)
self.assertIsNotNone(model_onnx)
dump_data_and_model(
X,
model,
model_onnx,
basename="SklearnLassoLarsIC-Dec4",
allow_failure="StrictVersion("
"onnxruntime.__version__)"
"<= StrictVersion('0.2.1')",
)
def cal_dep(Seq_dict, Deg_dict, miR_exp, mRNA_exp):
f = open('MicroTrans_results.txt', "w")
f.write('microRNA')
f.write('\t')
f.write('mRNA')
f.write('\t')
f.write('p-value')
f.write('\n')
#perform lasso regression
lasso_reg = {}
for (Key_mirna, Value_mrna) in Seq_dict.items():
print('processing microRNA ' + Key_mirna, end="\r")
TotalmRNA = []
for i in range(len(Value_mrna)):
TotalmRNA.append(mRNA_exp[Value_mrna[i]])
clf = linear_model.LassoLarsIC(criterion='bic')
clf.fit(np.transpose(np.asarray(TotalmRNA)),
np.asarray(miR_exp[Key_mirna]))
if len(np.nonzero(clf.coef_)) == 0:
continue
stdev = bootstrap(np.asarray(miR_exp[Key_mirna]),
np.asarray(TotalmRNA), len(Value_mrna))
for j in range(len(clf.coef_)):
if clf.coef_[j] != 0:
lasso_reg[(Key_mirna, Value_mrna[j])] = 1 - round(
scipy.stats.norm(0, 1).cdf(clf.coef_[j] / stdev[j]), 3)
lasso_reg_set = set(lasso_reg)
deg_set = set(Deg_dict)
sharedKey = {}
for inter_key in lasso_reg_set.intersection(deg_set):
sharedKey[(inter_key[0], inter_key[1])] = 1
Pvalue = scipy.stats.combine_pvalues(
[float(lasso_reg[inter_key]),
float(deg_set[inter_key])],
method='fisher')
output(inter_key[0], inter_key[1], Pvalue, f)
for uniq_key in lasso_reg.keys():
if uniq_key not in sharedKey.keys():
output(uniq_key[0], uniq_key[1], lasso_reg[uniq_key], f)
for uniq_key in Deg_dict.keys():
if uniq_key not in sharedKey.keys():
output(uniq_key[0], uniq_key[1], Deg_dict[uniq_key][0], f)
f.close()
print('Succesfully finished, the results are in MicroTrans_results.txt')
return None
def models(self) -> Dict[str, LinearModel]:
return {
"LinearRegression": linear_model.LinearRegression(
), # LinearRegression([…]) Ordinary least squares Linear Regression.
"RidgeCV": linear_model.RidgeCV(
cv=5
), # RidgeCV([alphas, …]) Ridge regression with built-in cross-validation.
"LassoLars": linear_model.LassoLars(
eps=0.01
), # LassoLars([alpha, …]) Lasso model fit with Least Angle Regression a.k.a.
"LassoLarsIC": linear_model.LassoLarsIC(
eps=0.01
), # LassoLarsIC([criterion, …]) Lasso model fit with Lars using BIC or AIC for model selection
"ARDRegression": linear_model.ARDRegression(
), # ARDRegression([n_iter, tol, …]) Bayesian ARD regression.
"ElasticNet": linear_model.ElasticNet(
), # linear_model.ElasticNet([alpha, l1_ratio, …]) Linear regression with combined L1 and L2 priors as regularizer.
}
def _estimate_model(self):
"""Estimates lasso regression object.
Returns
-------
model : sklearn lasso regression or lasso cv object
Fitted lasso model.
"""
###Lars Algorithm
if self.solver == "Lars":
self.underlying = linear_model.LassoLars(
fit_intercept=self.intercept, normalize=False)
if self.cv_folds is 'IC': #For AIC/BIC. criterion kwarg should be provided.
model = linear_model.LassoLarsIC(fit_intercept=self.intercept,
normalize=False,
**self.kwargs)
elif self.cv_folds is not None:
model = linear_model.LassoLarsCV(fit_intercept=self.intercept,
cv=self.cv_folds,
normalize=False,
**self.kwargs)
else:
model = linear_model.Lasso(fit_intercept=self.intercept,
**self.kwargs)
###Coordinate Descent Algorithm
elif self.solver == "Coordinate Descent":
self.underlying = linear_model.Lasso(fit_intercept=self.intercept)
if self.cv_folds is not None:
model = linear_model.LassoCV(fit_intercept=self.intercept,
cv=self.cv_folds,
**self.kwargs)
else:
model = linear_model.Lasso(fit_intercept=self.intercept,
**self.kwargs)
else:
raise NotImplementedError(
'Solver not implemented. Choices are Lars or Coordinate Descent.'
)
#self.model.fit(np.asanyarray(self.x_train.values,order='F'), self.y_train)
model.fit(self.x_train, self.y_train)
return model
def test_LassoCV(self, criterion):
diabetes = datasets.load_diabetes()
X = diabetes.data
y = diabetes.target
X = pp.normalize(X)
df = pdml.ModelFrame(diabetes)
df.data = df.data.pp.normalize()
mod1 = lm.LassoLarsIC(criterion=criterion)
mod1.fit(X, y)
mod2 = df.lm.LassoLarsIC(criterion=criterion)
df.fit(mod2)
self.assertAlmostEqual(mod1.alpha_, mod2.alpha_)
expected = mod1.predict(X)
predicted = df.predict(mod2)
self.assertIsInstance(predicted, pdml.ModelSeries)
self.assert_numpy_array_almost_equal(predicted.values, expected)
print("Set up KFolds...")
n_splits = 5
kf = KFold(n_splits=n_splits)
kf.get_n_splits(X)
predictions0 = np.zeros((test.shape[0], n_splits))
predictions1 = np.zeros((test.shape[0], n_splits))
score = 0
print("Starting ", n_splits, "-fold CV loop...")
oof_predictions = np.zeros(X.shape[0])
for fold, (train_index, test_index) in enumerate(kf.split(X)):
X_train, X_valid = X[train_index, :], X[test_index, :]
y_train, y_valid = y[train_index], y[test_index]
clf = linear_model.LassoLarsIC()
clf.fit(X_train, y_train)
pred0 = clf.predict(X)
pred1 = clf.predict(test)
oof_predictions[test_index] = clf.predict(X_valid)
predictions0[:, fold] = pred0
predictions1[:, fold] = pred1
score += r2_score(y_train, clf.predict(X_train))
print('Fold %d: Score %f' % (fold, clf.score(X_train, y_train)))
prediction0 = predictions0.mean(axis=1)
prediction1 = predictions1.mean(axis=1)
score /= n_splits
oof_score = r2_score(y, oof_predictions)
def run_models(fullname, choose, delete):
folder, filename = os.path.split(fullname)
folder = folder.replace('/uploads', '/static')
a = choose
b = delete
df = pd.read_csv(fullname)
y_data = np.array(df.loc[:, a])
X_picked = df.drop(a, 1)
for row in b:
X_picked = X_picked.drop(row, 1)
num_column = len(X_picked.columns.values)
imp = Imputer()
X_picked = imp.fit_transform(X_picked)
y_data = imp.fit_transform(y_data)
y_data = np.reshape(y_data, (len(df), ))
X_picked = preprocessing.minmax_scale(X_picked, feature_range=(0, 1))
y_data = preprocessing.minmax_scale(y_data, feature_range=(0, 1))
def add_layer(x, in_size, out_size, activation_function, layer, dropout):
num_layer = 'layer%s' % layer
with tf.name_scope(num_layer):
global W
tf.set_random_seed(666)
W = tf.Variable(tf.random_uniform([in_size, out_size]))
with tf.name_scope('y'):
output = tf.matmul(x, W)
if activation_function is 1:
out = tf.nn.softsign(output)
elif activation_function is 0:
out = tf.nn.dropout(output, dropout)
else:
out = output
return out
# Create input variable
with tf.name_scope('Input'):
x = tf.placeholder(tf.float32, [None, num_column])
y_ = tf.placeholder(tf.float32, [None, 1])
keep_prob = tf.placeholder(tf.float32) #dropout probability
dropout = 0.85
# Build up layers
with tf.name_scope('Layer'):
y = add_layer(x, num_column, 1, 2, 1, dropout)
# input
header = list(df.columns.values)
list_header = []
for i in range(len(header)):
if header[i] != a:
list_header.append(header[i])
# cost functiontensoflow
with tf.name_scope('Cost'):
cost = tf.reduce_mean(tf.square(y_ - y))
# training
Training = tf.train.AdamOptimizer(0.01)
train = Training.minimize(cost)
with tf.Session() as sess:
sess = tf.Session()
init = tf.initialize_all_variables()
sess.run(init)
feed = {}
chi_sum = 0
predict = []
actual = []
chi_plot = []
i_count = []
counter = 0
df1 = np.float32(df)
for i in xrange(len(df.values)):
xs = np.array(np.float32([X_picked[i]]))
ys = np.array(np.float32([[y_data[i]]]))
feed = {x: xs, y_: ys, keep_prob: dropout}
prediction = sess.run(y, feed_dict=feed)
sess.run(train, feed_dict=feed)
chi = (np.square(ys - prediction)) / (prediction)
chi_sum += chi
predict.append(float(prediction) * 1000)
actual.append(float(ys) * 1000)
chi_plot.append(float(chi))
i_count.append(i)
weight = np.array(sess.run(W)).flatten()
r2_score_NN = r2_score(actual, predict)
# scikit learn machine learning
# Ridge Regression
model1 = Ridge(alpha=1.0)
model1.fit(X_picked, y_data)
prediction_ridge = cross_val_predict(model1, X_picked, y_data, cv=10)
r2_score_LR = r2_score(y_data, prediction_ridge)
r_linear_ridge = stats.pearsonr(prediction_ridge, y_data)
prediction_ridge = np.array(prediction_ridge) * 1000
ridge_coef = model1.coef_
# Lasso Regression
clf = linear_model.LassoLarsIC(criterion='aic')
clf.fit(X_picked, y_data)
prediction_lasso = cross_val_predict(clf, X_picked, y_data, cv=10)
r2_score_lasso_aic = r2_score(y_data, prediction_lasso)
r_linear_lassoAIC = stats.pearsonr(prediction_lasso, y_data)
prediction_lasso = np.array(prediction_lasso) * 1000
lasso_coef = clf.coef_
# Stats and Plot
c = stats.pearsonr(actual, predict)
plt.figure()
plt.scatter(predict, actual, s=3)
plt.xlabel('Prediction')
plt.ylabel('Observation')
plt.title('This is Neural Network')
uuid = hashlib.md5(filename + '_NN:' + a).hexdigest()
# uuid = str(uuid.uuid4())
id_nn = uuid
filename = os.path.join(folder, 'images/%s.png' % id_nn)
plt.savefig(filename)
plt.figure()
plt.scatter(prediction_ridge, actual, s=3)
plt.xlabel('Prediction')
plt.ylabel('Observation')
plt.title('This is Ridge Regression')
uuid = hashlib.md5(filename + '_RIDGE:' + a).hexdigest()
id_ridge = uuid
filename1 = os.path.join(folder, 'images/%s.png' % id_ridge)
plt.savefig(filename1)
plt.figure()
plt.scatter(prediction_lasso, actual, s=3)
plt.xlabel('Prediction')
plt.ylabel('Observation')
plt.title('This is Lasso Regression')
uuid = hashlib.md5(filename + '_LASSO:' + a).hexdigest()
id_lasso = uuid
filename2 = os.path.join(folder, 'images/%s.png' % id_lasso)
plt.savefig(filename2)
return dict(a=a,
id_nn=id_nn,
id_ridge=id_ridge,
id_lasso=id_lasso,
ridge_coef=ridge_coef,
lasso_coef=lasso_coef,
weight=weight,
r_linear_lassoAIC=r_linear_lassoAIC,
r_linear_ridge=r_linear_ridge,
r2_score_NN=r2_score_NN,
uuid=uuid)
def __init__(
self,
method,
yrange,
params,
i=0
): #TODO: yrange doesn't currently do anything. Remove or do something with it!
self.algorithm_list = [
'PLS',
'GP',
'OLS',
'OMP',
'Lasso',
'Elastic Net',
'Ridge',
'Bayesian Ridge',
'ARD',
'LARS',
'LASSO LARS',
'SVR',
'KRR',
]
self.method = method
self.outliers = None
self.ransac = False
print(params)
if self.method[i] == 'PLS':
self.model = PLSRegression(**params[i])
if self.method[i] == 'OLS':
self.model = linear.LinearRegression(**params[i])
if self.method[i] == 'OMP':
# check whether to do CV or not
self.do_cv = params[i]['CV']
# create a temporary set of parameters
params_temp = copy.copy(params[i])
# Remove CV parameter
params_temp.pop('CV')
if self.do_cv is False:
self.model = linear.OrthogonalMatchingPursuit(**params_temp)
else:
params_temp.pop('precompute')
self.model = linear.OrthogonalMatchingPursuitCV(**params_temp)
if self.method[i] == 'LASSO':
# create a temporary set of parameters
params_temp = copy.copy(params[i])
# check whether to do CV or not
try:
self.do_cv = params[i]['CV']
# Remove CV parameter
params_temp.pop('CV')
except:
self.do_cv = False
if self.do_cv is False:
self.model = linear.Lasso(**params_temp)
else:
params_temp.pop('alpha')
self.model = linear.LassoCV(**params_temp)
if self.method[i] == 'Elastic Net':
params_temp = copy.copy(params[i])
try:
self.do_cv = params[i]['CV']
params_temp.pop('CV')
except:
self.do_cv = False
if self.do_cv is False:
self.model = linear.ElasticNet(**params_temp)
else:
params_temp['l1_ratio'] = [.1, .5, .7, .9, .95, .99, 1]
self.model = linear.ElasticNetCV(**params_temp)
if self.method[i] == 'Ridge':
# create a temporary set of parameters
params_temp = copy.copy(params[i])
try:
# check whether to do CV or not
self.do_cv = params[i]['CV']
# Remove CV parameter
params_temp.pop('CV')
except:
self.do_cv = False
if self.do_cv:
self.model = linear.RidgeCV(**params_temp)
else:
self.model = linear.Ridge(**params_temp)
if self.method[i] == 'BRR':
self.model = linear.BayesianRidge(**params[i])
if self.method[i] == 'ARD':
self.model = linear.ARDRegression(**params[i])
if self.method[i] == 'LARS':
# create a temporary set of parameters
params_temp = copy.copy(params[i])
try:
# check whether to do CV or not
self.do_cv = params[i]['CV']
# Remove CV parameter
params_temp.pop('CV')
except:
self.do_cv = False
if self.do_cv is False:
self.model = linear.Lars(**params_temp)
else:
self.model = linear.LarsCV(**params_temp)
if self.method[i] == 'LASSO LARS':
model = params[i]['model']
params_temp = copy.copy(params[i])
params_temp.pop('model')
if model == 0:
self.model = linear.LassoLars(**params_temp)
elif model == 1:
self.model = linear.LassoLarsCV(**params_temp)
elif model == 2:
self.model = linear.LassoLarsIC(**params_temp)
else:
print("Something went wrong, \'model\' should be 0, 1, or 2")
if self.method[i] == 'SVR':
self.model = svm.SVR(**params[i])
if self.method[i] == 'KRR':
self.model = kernel_ridge.KernelRidge(**params[i])
if self.method[i] == 'GP':
# get the method for dimensionality reduction and the number of components
self.reduce_dim = params[i]['reduce_dim']
self.n_components = params[i]['n_components']
# create a temporary set of parameters
params_temp = copy.copy(params[i])
# Remove parameters not accepted by Gaussian Process
params_temp.pop('reduce_dim')
params_temp.pop('n_components')
self.model = GaussianProcess(**params_temp)
def model_selection():
# This is to avoid division by zero while doing np.log10
EPSILON = 1e-5
# #############################################################################
# LassoLarsIC: least angle regression with BIC/AIC criterion
model_bic = linear_model.LassoLarsIC(criterion='bic')
model_bic.fit(data, label)
# alpha_bic_ = model_bic.alpha_
model_aic = linear_model.LassoLarsIC(criterion='aic')
model_aic.fit(data, label)
# alpha_aic_ = model_aic.alpha_
plt.figure()
plot_ic_criterion(model_aic, 'AIC', 'b', EPSILON)
plot_ic_criterion(model_bic, 'BIC', 'r', EPSILON)
plt.legend()
plt.title('Information-criterion for model selection')
plt.savefig('information_criterion_model_selection.png')
# #############################################################################
# LassoCV: coordinate descent
# Compute paths
model = linear_model.LassoCV(cv=10).fit(data, label)
# Display results
m_log_alphas = -np.log10(model.alphas_ + EPSILON)
plt.figure()
ymin, ymax = 20, 300
plt.plot(m_log_alphas, model.mse_path_, ':')
plt.plot(m_log_alphas,
model.mse_path_.mean(axis=-1),
'k',
label='Average across the folds',
linewidth=2)
plt.axvline(-np.log10(model.alpha_ + EPSILON),
linestyle='--',
color='k',
label='alpha: CV estimate')
plt.legend()
plt.xlabel('-log(alpha)')
plt.ylabel('Mean square error')
plt.title('Mean square error on each fold: coordinate descent ')
plt.axis('tight')
plt.ylim(ymin, ymax)
plt.savefig('lasso_model_selection.png')
# #############################################################################
# LassoLarsCV: least angle regression
# Compute paths
model = linear_model.LassoLarsCV(cv=10).fit(data, label)
# Display results
m_log_alphas = -np.log10(model.cv_alphas_ + EPSILON)
plt.figure()
plt.plot(m_log_alphas, model.mse_path_, ':')
plt.plot(m_log_alphas,
model.mse_path_.mean(axis=-1),
'k',
label='Average across the folds',
linewidth=2)
plt.axvline(-np.log10(model.alpha_),
linestyle='--',
color='k',
label='alpha CV')
plt.legend()
plt.xlabel('-log(alpha)')
plt.ylabel('Mean square error')
plt.title('Mean square error on each fold: Lars')
plt.axis('tight')
plt.ylim(ymin, ymax)
plt.savefig('lasso_Lars_model_selection.png')
from itertools import cycle
from sklearn import svm, datasets
from sklearn.metrics import roc_curve, auc, precision_recall_curve, average_precision_score
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import label_binarize
from sklearn.multiclass import OneVsRestClassifier
from scipy import interp
from sklearn.decomposition import PCA
# Problem 1
dataset = pd.read_csv("ecs171.dataset.txt", sep='\s+', header = None).values
dataset_x = dataset[1:,7:4502].astype(float)
dataset_y = dataset[1:,5].astype(float)
# Find optimal alpha using bic criteria
bic_model = linear_model.LassoLarsIC(criterion = 'bic')
bic_model.fit(dataset_x, dataset_y)
optimal_alpha = bic_model.alpha_
clf = linear_model.Lasso(alpha = optimal_alpha)
clf.fit(dataset_x, dataset_y)
#count the number of non-zero features
count = 0
for feature in clf.coef_:
if feature != 0:
count += 1
print("The number of non-zero features is", count)
# Cross validation
cv_model = linear_model.LassoLarsCV(cv = 10).fit(dataset_x, dataset_y)
print("Ten fold cross-validation error is", cv_model.cv_mse_path_.mean())
# Problem 2
], n_jobs=-1),
),
])
X, y = make_xy_data('./data/merged_data.csv', ['surface_m2', 'piece'])
X_train, X_test, y_train, y_test = train_test_split(X, y,
test_size=0.2,
random_state=2)
X_tr = features.fit_transform(X_train, None)
X_te = features.transform(X_test)
###############################################################################
model_bic = lm.LassoLarsIC(criterion='bic', verbose=2)
t1 = time.time()
model_bic.fit(X_tr, y_train)
t_bic = time.time() - t1
alpha_bic_ = model_bic.alpha_
model_aic = lm.LassoLarsIC(criterion='aic', verbose=2)
model_aic.fit(X_tr, y_train)
alpha_aic_ = model_aic.alpha_
plt.figure()
plot_ic_criterion(model_aic, 'AIC', 'b')
plot_ic_criterion(model_bic, 'BIC', 'r')
plt.legend()
plt.title('Information-criterion for model selection (training time %.3fs)'
% t_bic)
classification_binary(svm.NuSVC(kernel="rbf", **SVC_PARAMS)),
classification(svm.SVC(kernel="rbf", **SVC_PARAMS)),
classification(svm.NuSVC(kernel="rbf", **SVC_PARAMS)),
# Linear Regression
regression(linear_model.LinearRegression()),
regression(linear_model.HuberRegressor()),
regression(linear_model.ElasticNet(random_state=RANDOM_SEED)),
regression(linear_model.ElasticNetCV(random_state=RANDOM_SEED)),
regression(linear_model.TheilSenRegressor(random_state=RANDOM_SEED)),
regression(linear_model.Lars()),
regression(linear_model.LarsCV()),
regression(linear_model.Lasso(random_state=RANDOM_SEED)),
regression(linear_model.LassoCV(random_state=RANDOM_SEED)),
regression(linear_model.LassoLars()),
regression(linear_model.LassoLarsIC()),
regression(linear_model.OrthogonalMatchingPursuit()),
regression(linear_model.OrthogonalMatchingPursuitCV()),
regression(linear_model.Ridge(random_state=RANDOM_SEED)),
regression(linear_model.RidgeCV()),
regression(linear_model.BayesianRidge()),
regression(linear_model.ARDRegression()),
regression(linear_model.SGDRegressor(random_state=RANDOM_SEED)),
regression(
linear_model.PassiveAggressiveRegressor(random_state=RANDOM_SEED)),
# Logistic Regression
classification(
linear_model.LogisticRegression(random_state=RANDOM_SEED)),
classification(
linear_model.LogisticRegressionCV(random_state=RANDOM_SEED)),
rowitems = line.rstrip('\n').split('\t')
X_item = map(float, rowitems[0:-2])
y_item = float(rowitems[-2])
sid_item = int(rowitems[-1])
X.append(X_item)
y.append(y_item)
sid.append(sid_item)
count += 1
# print count
X_train = np.array(X)
y_train = np.array(y)
sid_train = np.array(sid)
# regr = linear_model.LinearRegression() # this runs linear regression
regr = linear_model.LassoLarsIC(
criterion='bic', fit_intercept=False) # this runs lasso regression
regr.fit(X_train, y_train)
fit_coef = regr.coef_
# P # for GLM, P is diagonal variance matrix estimated using fit_coef_
fit_covinv = np.dot(
X_train.T,
X_train) # for GLM, use np.dot(np.dot(X_train.T, P), X_train) instead
fit_lambda = regr.alpha_
A = np.dot(inv(fit_covinv), X_train.T)
e = y - np.dot(
X_train, fit_coef
) # for GLM, transform np.dot(X_train, fit_coef) using link function
betahat_c = fit_coef + np.dot(A, e)
spca = dcm.SparsePCA()
sparse_alpha_opts = [0.1, 0.5, 1, 2, 5, 10]
kpca = dcm.KernelPCA()
kernel_opts = ["linear", "rbf", "sigmoid"]
n_component_opts = [0.7, 0.8, 0.9, 0.95, 0.99]
lasso = linear_model.LassoCV(max_iter=100000,
n_jobs=28,
alphas=np.arange(0.1, 50, 0.1))
lasso_lars = linear_model.LassoLarsCV(n_jobs=28,
max_iter=100000,
max_n_alphas=10000)
eps_opts = [10.0, 5.0, 2.0, 1.5, 0.9, 0.1, 0.01, 0.001, 0.0001]
elastic = linear_model.ElasticNetCV(alphas=np.arange(0.1, 50, 0.1),
max_iter=100000)
l1_ratio_opts = [0.1, 0.5, 0.9, 0.95, 0.99]
lasso_lars_bay = linear_model.LassoLarsIC(max_iter=100000)
cv_opts = [5, 10, 15, 20]
param_dict = [
{
'clf': (lasso, ),
'pca': (spca, ),
'pca__alpha': sparse_alpha_opts,
'clf__selection': ['random', 'cyclic'],
'clf__cv': cv_opts,
},
{
'clf': (lasso, ),
'pca': (kpca, ),
'pca__kernel': kernel_opts,
'clf__selection': ['random', 'cyclic'],
def fit_regression(P, x, u, rule="LS", retall=False, **kws):
"""
Fit a polynomial chaos expansion using linear regression.
Parameters
----------
P : Poly
Polynomial chaos expansion with `P.shape=(M,)` and `P.dim=D`.
x : array_like
Collocation nodes with `x.shape=(D,K)`.
u : array_like
Model evaluations with `len(u)=K`.
retall : bool
If True return uhat in addition to R
rule : str
Regression method used.
The follwong methods uses scikits-learn as backend.
See `sklearn.linear_model` for more details.
Key Scikit-learn Description
--- ------------ -----------
Parameters Description
---------- -----------
"BARD" ARDRegression Bayesian ARD Regression
n_iter=300 Maximum iterations
tol=1e-3 Optimization tolerance
alpha_1=1e-6 Gamma scale parameter
alpha_2=1e-6 Gamma inverse scale parameter
lambda_1=1e-6 Gamma shape parameter
lambda_2=1e-6 Gamma inverse scale parameter
threshold_lambda=1e-4 Upper pruning threshold
"BR" BayesianRidge Bayesian Ridge Regression
n_iter=300 Maximum iterations
tol=1e-3 Optimization tolerance
alpha_1=1e-6 Gamma scale parameter
alpha_2=1e-6 Gamma inverse scale parameter
lambda_1=1e-6 Gamma shape parameter
lambda_2=1e-6 Gamma inverse scale parameter
"EN" ElastiNet Elastic Net
alpha=1.0 Dampening parameter
rho Mixing parameter in [0,1]
max_iter=300 Maximum iterations
tol Optimization tolerance
"ENC" ElasticNetCV EN w/Cross Validation
rho Dampening parameter(s)
eps=1e-3 min(alpha)/max(alpha)
n_alphas Number of alphas
alphas List of alphas
max_iter Maximum iterations
tol Optimization tolerance
cv=3 Cross validation folds
"LA" Lars Least Angle Regression
n_nonzero_coefs Number of non-zero coefficients
eps Cholesky regularization
"LAC" LarsCV LAR w/Cross Validation
max_iter Maximum iterations
cv=5 Cross validation folds
max_n_alphas Max points for residuals in cv
"LAS" Lasso Least Absolute Shrinkage and
Selection Operator
alpha=1.0 Dampening parameter
max_iter Maximum iterations
tol Optimization tolerance
"LASC" LassoCV LAS w/Cross Validation
eps=1e-3 min(alpha)/max(alpha)
n_alphas Number of alphas
alphas List of alphas
max_iter Maximum iterations
tol Optimization tolerance
cv=3 Cross validation folds
"LL" LassoLars Lasso and Lars model
max_iter Maximum iterations
eps Cholesky regularization
"LLC" LassoLarsCV LL w/Cross Validation
max_iter Maximum iterations
cv=5 Cross validation folds
max_n_alphas Max points for residuals in cv
eps Cholesky regularization
"LLIC" LassoLarsIC LL w/AIC or BIC
criterion "AIC" or "BIC" criterion
max_iter Maximum iterations
eps Cholesky regularization
"OMP" OrthogonalMatchingPursuit
n_nonzero_coefs Number of non-zero coefficients
tol Max residual norm (instead of non-zero coef)
Local methods
Key Description
--- -----------
"LS" Ordenary Least Squares
"T" Ridge Regression/Tikhonov Regularization
order Order of regularization (or custom matrix)
alpha Dampning parameter (else estimated from gcv)
"TC" T w/Cross Validation
order Order of regularization (or custom matrix)
alpha Dampning parameter (else estimated from gcv)
Returns
-------
R[, uhat]
R : Poly
Fitted polynomial with `R.shape=u.shape[1:]` and `R.dim=D`.
uhat : np.ndarray
The Fourier coefficients in the estimation.
Examples
--------
>>> P = cp.Poly([1, x, y])
>>> s = [[-1,-1,1,1], [-1,1,-1,1]]
>>> u = [0,1,1,2]
>>> print fit_regression(P, s, u)
0.5q1+0.5q0+1.0
"""
x = np.array(x)
if len(x.shape) == 1:
x = x.reshape(1, *x.shape)
u = np.array(u)
Q = P(*x).T
shape = u.shape[1:]
u = u.reshape(u.shape[0], int(np.prod(u.shape[1:])))
rule = rule.upper()
# Local rules
if rule == "LS":
uhat = la.lstsq(Q, u)[0].T
elif rule == "T":
uhat, alphas = rlstsq(Q, u, kws.get("order", 0),
kws.get("alpha", None), False, True)
uhat = uhat.T
elif rule == "TC":
uhat = rlstsq(Q, u, kws.get("order", 0), kws.get("alpha", None), True)
uhat = uhat.T
else:
# Scikit-learn wrapper
try:
_ = lm
except:
raise NotImplementedError("sklearn not installed")
if rule == "BARD":
solver = lm.ARDRegression(fit_intercept=False, copy_X=False, **kws)
elif rule == "BR":
kws["fit_intercept"] = kws.get("fit_intercept", False)
solver = lm.BayesianRidge(**kws)
elif rule == "EN":
kws["fit_intercept"] = kws.get("fit_intercept", False)
solver = lm.ElasticNet(**kws)
elif rule == "ENC":
kws["fit_intercept"] = kws.get("fit_intercept", False)
solver = lm.ElasticNetCV(**kws)
elif rule == "LA": # success
kws["fit_intercept"] = kws.get("fit_intercept", False)
solver = lm.Lars(**kws)
elif rule == "LAC":
kws["fit_intercept"] = kws.get("fit_intercept", False)
solver = lm.LarsCV(**kws)
elif rule == "LAS":
kws["fit_intercept"] = kws.get("fit_intercept", False)
solver = lm.Lasso(**kws)
elif rule == "LASC":
kws["fit_intercept"] = kws.get("fit_intercept", False)
solver = lm.LassoCV(**kws)
elif rule == "LL":
kws["fit_intercept"] = kws.get("fit_intercept", False)
solver = lm.LassoLars(**kws)
elif rule == "LLC":
kws["fit_intercept"] = kws.get("fit_intercept", False)
solver = lm.LassoLarsCV(**kws)
elif rule == "LLIC":
kws["fit_intercept"] = kws.get("fit_intercept", False)
solver = lm.LassoLarsIC(**kws)
elif rule == "OMP":
solver = lm.OrthogonalMatchingPursuit(**kws)
uhat = solver.fit(Q, u).coef_
u = u.reshape(u.shape[0], *shape)
R = po.sum((P * uhat), -1)
R = po.reshape(R, shape)
if retall == 1:
return R, uhat
elif retall == 2:
if rule == "T":
return R, uhat, Q, alphas
return R, uhat, Q
return R
x_train, x_test, y_train, y_test = train_test_split(X,
Y,
test_size=0.33,
random_state=1)
alphas = np.logspace(-4, 0.5, 30)
models = {
'LinReg': linear_model.LinearRegression(),
'Ridge': linear_model.Ridge(),
'RidgeCV': linear_model.RidgeCV(),
'Lasso': linear_model.Lasso(),
'LassoCV': linear_model.LassoCV(),
# 'LassoLarsCV': linear_model.LassoLarsCV(),
'LassoLarsIC_AIC': linear_model.LassoLarsIC(criterion='aic'),
'LassoLarsIC_BIC': linear_model.LassoLarsIC(criterion='bic'),
'LassoLars': linear_model.LassoLars(),
'ElasticNet': linear_model.ElasticNet()
}
scores = {}
t_check = []
t2 = 0
for name, model in models.items():
for alpha in alphas:
model.alpha = alpha
if 'CV' in name:
model.cv = 20
#check time of training
t1 = time.time()
def lasso_predict(df):
K=0.65306122
a_list=list(df.columns)
names = [item for item in a_list if item not in inter_col]
xdata=df[names]
labels=np.array(df.REVENUE[1:],dtype=np.float64)
xList=np.array(xdata[:-1],dtype=np.float64)
X,Y=normalize(xList,labels)
X[np.isnan(X)]=0
#give value of cv
n=len(df)
if n>8:
cv=8
else:
cv=n-2
#Call LassoCV from sklearn.linear_model
X=np.nan_to_num(X)
Rev_Model = LassoLarsCV(cv=cv).fit(X, Y)
alphas, coefs, _ = linear_model.lasso_path(X, Y, return_models=False)
nattr, nalpha = coefs.shape
#find coefficient ordering
nzList = []
for iAlpha in range(1,nalpha):
coefList = list(coefs[: ,iAlpha])
nzCoef = [index for index in range(nattr) if coefList[index] != 0.0]
for q in nzCoef:
if not(q in nzList):
nzList.append(q)
#find coefficients corresponding to best alpha value. alpha value corresponding to
#normalized X and normalized Y is Rev_Model.alpha_
alphaStar =Rev_Model.alpha_
indexLTalphaStar = [index for index in range(100) if alphas[index] > alphaStar]
indexStar = max(indexLTalphaStar)
#here's the set of coefficients to deploy
coefStar = list(coefs[:,indexStar])
#The coefficients on normalized attributes give another slightly different ordering
absCoef = [abs(a) for a in coefStar]
#sort by magnitude
coefSorted = sorted(absCoef, reverse=True)
idxCoefSize = [absCoef.index(a) for a in coefSorted if not(a == 0.0)]
vari_nm= [xdata.columns[idxCoefSize[i]] for i in range(len(idxCoefSize))]
#use variables in vari_nmto regress
feat=min(len(vari_nm),int(K*len(df)))
vari_nm=vari_nm[:feat]
y=np.array(xdata.REVENUE[1:],dtype=np.float64)
x=np.array(xdata[vari_nm][:-1],dtype=np.float64)
xpred=np.array(xdata[vari_nm][-2:])
rev_q1_true=np.float(xdata.REVENUE[-1:])/1000000
X1,Y1=normalize(x,y)
reg = linear_model.LassoLarsIC(criterion='aic')
reg.fit(X1,Y1)
coefs=reg.coef_
score=reg.score(X1,Y1)
vari_nm1=[vari_nm[i] for i in range(len(vari_nm)) if coefs[i]!=0]
if (len(vari_nm1)>1)&(score>0.412626):
x=np.array(xdata[vari_nm1][:-1],dtype=np.float64)
xpred=np.array(xdata[vari_nm1][-2:])
linreg=LinearRegression()
linreg.fit(x, y)
score=linreg.score(x,y)
rev_q1=linreg.predict(xpred)[0]/1000000
rev_q2=linreg.predict(xpred)[1]/1000000+rev_q1_true
return [rev_q1_true,rev_q1,rev_q2]
