def explain_node(self, node_idx, x, edge_index, **kwargs):
probas = self.__init_predict__(x, edge_index, **kwargs)
x, probas, _, _, _, _ = self.__subgraph__(node_idx, x, probas,
edge_index, **kwargs)
x = x.detach().cpu().numpy() # (n, d)
y = probas.detach().cpu().numpy() # (n, classes)
n, d = x.shape
K = self.__compute_kernel__(x, reduce=False) # (n, n, d)
L = self.__compute_kernel__(y, reduce=True) # (n, n, 1)
K_bar = self.__compute_gram_matrix__(K) # (n, n, d)
L_bar = self.__compute_gram_matrix__(L) # (n, n, 1)
K_bar = K_bar.reshape(n**2, d) # (n ** 2, d)
L_bar = L_bar.reshape(n**2, ) # (n ** 2,)
solver = LassoLars(self.rho,
fit_intercept=False,
normalize=False,
positive=True)
solver.fit(K_bar * n, L_bar * n)
return solver.coef_
def RunLARSScikit():
totalTimer = Timer()
# Load input dataset.
Log.Info("Loading dataset", self.verbose)
inputData = np.genfromtxt(self.dataset[0], delimiter=',')
responsesData = np.genfromtxt(self.dataset[1], delimiter=',')
opts = {}
if "lambda1" in options:
opts["alpha"] = float(options.pop("lambda1"))
if "max_iterations" in options:
opts["max_iter"] = int(options.pop("max_iterations"))
if "epsilon" in options:
opts["eps"] = float(options.pop("epsilon"))
if len(options) > 0:
Log.Fatal("Unknown parameters: " + str(options))
raise Exception("unknown parameters")
try:
with totalTimer:
# Perform LARS.
model = LassoLars(**opts)
model.fit(inputData, responsesData)
out = model.coef_
except Exception as e:
return -1
return totalTimer.ElapsedTime()
def RunLARSScikit(q):
totalTimer = Timer()
# Load input dataset.
Log.Info("Loading dataset", self.verbose)
inputData = np.genfromtxt(self.dataset[0], delimiter=',')
responsesData = np.genfromtxt(self.dataset[1], delimiter=',')
try:
with totalTimer:
# Get all the parameters.
lambda1 = re.search("-l (\d+)", options)
lambda1 = 0.0 if not lambda1 else int(lambda1.group(1))
# Perform LARS.
model = LassoLars(alpha=lambda1)
model.fit(inputData, responsesData)
out = model.coef_
except Exception as e:
q.put(-1)
return -1
time = totalTimer.ElapsedTime()
q.put(time)
return time
def cv_train_lasso_lars_with_sparse_refit(x_train,
y_train,
pval_cutoff=0.001,
do_sparse_refit=True):
model = LassoLarsCV(n_jobs=-1, cv=min(x_train.shape[0], 10))
model.fit(x_train, y_train)
best_alpha_idx = int(np.argwhere(model.alpha_ == model.cv_alphas_))
if do_sparse_refit:
sparse_alpha_idx = -1
for i in range(best_alpha_idx + 1, len(model.cv_alphas_)):
pval = ttest_ind(model.mse_path_[best_alpha_idx],
model.mse_path_[i]).pvalue
if pval < pval_cutoff:
sparse_alpha_idx = i - 1
break
if sparse_alpha_idx == -1:
# take the sparsest solution
sparse_alpha_idx = len(model.cv_alphas_) - 1
model_sparse = LassoLars(alpha=model.cv_alphas_[sparse_alpha_idx])
model_sparse.fit(x_train, y_train)
return model_sparse
else:
return model
def LassoLars_score(X,y,**l1_parameters):
"""
Score predictor based on `scikit-learn`_ LassoLars regression.
Args:
X (pandas.DataFrame): Transcriptor factor gene expressions where rows
are experimental conditions and columns are transcription factors
y (pandas.Series): Target gene expression vector where rows are
experimental conditions
**l1_parameters: Named parameters for sklearn Lasso regression
Returns:
numpy.array: co-regulation scores.
The i-th element of the score array represents the score assigned by the
sklearn LassoLars regressor to the regulatory relationship between the
target gene and transcription factor i.
Examples:
>>> import pandas as pd
>>> import numpy as np
>>> np.random.seed(0)
>>> tfs = pd.DataFrame(np.random.randn(5,3),
index =["c1","c2","c3","c4","c5"],
columns=["tf1","tf2","tf3"])
>>> tg = pd.Series(np.random.randn(5),index=["c1","c2","c3","c4","c5"])
>>> scores = LassoLars_score(tfs,tg, alpha=0.01)
>>> scores
array([0.12179406, 0.92205553, 0.15503451])
"""
regressor = LassoLars(**l1_parameters)
regressor.fit(X, y)
scores = np.abs(regressor.coef_)
return(scores)
def lassolarsdimension(data, label):
lassolarscv = LassoLarsCV(cv=5, max_iter=400).fit(data, label)
lassolars = LassoLars(alpha=lassolarscv.alpha_) #生成LassoLars对象
x_lassolars = lassolars.fit(data, label)
mask = x_lassolars.coef_ != 0
new_data = data[:, mask]
return new_data, mask
def online_dict_learning(X, lmda, D_0, T, k_cluster, eps, _NF=200):
'''
algo 1 in the paper
D_0: R^(m * k)
X: R^(n * m)
'''
n_dim, m_dim = X.shape
A_t = np.zeros((k_cluster, k_cluster))
B_t = np.zeros((m_dim, k_cluster))
D_t = D_0
t_start = time.time()
# print(lmda, _NF, eps)
for t in range(T):
# t_start_online = time.time()
sample_idx = np.random.randint(0, n_dim)
x_sample = X[sample_idx, :]
lars_lasso = LassoLars(alpha=lmda)
lars_lasso.fit(D_t, x_sample)
alpha_t = lars_lasso.coef_
A_t += np.matmul(alpha_t.reshape(k_cluster, 1),
alpha_t.reshape(1, k_cluster))
B_t += np.matmul(x_sample.reshape(m_dim, 1),
alpha_t.reshape(1, k_cluster))
D_t = dict_update(D_t, A_t, B_t, eps=eps, _NF=_NF)
# print('===== Iteration in online dictionary learning cost {:.04f}s'.format(time.time() - t_start_online))
print('Dcitionary update done! Time elapse {:.04f}s'.format(time.time() -
t_start))
return D_t
def RunLARSScikit():
totalTimer = Timer()
# Load input dataset.
Log.Info("Loading dataset", self.verbose)
inputData = np.genfromtxt(self.dataset[0], delimiter=',')
responsesData = np.genfromtxt(self.dataset[1], delimiter=',')
opts = {}
if "lambda1" in options:
opts["alpha"] = float(options.pop("lambda1"))
if "max_iterations" in options:
opts["max_iter"] = int(options.pop("max_iterations"))
if "epsilon" in options:
opts["eps"] = float(options.pop("epsilon"))
if len(options) > 0:
Log.Fatal("Unknown parameters: " + str(options))
raise Exception("unknown parameters")
try:
with totalTimer:
# Perform LARS.
model = LassoLars(**opts)
model.fit(inputData, responsesData)
out = model.coef_
except Exception as e:
return -1
return totalTimer.ElapsedTime()
def Lasso_fit(alpha, x, y):
solver = LassoLars(alpha=alpha, fit_intercept=False, max_iter=3000)
solver.alpha = alpha
solver.fit(x, y)
idxs = solver.coef_ != 0.
c_cal = sum(idxs)
return idxs, c_cal
def __init__(
self,
propensity_learner=None,
learner=None,
learner_c=None,
learner_t=None,
delta=0.001,
):
"""Setup a DoubleRobustEstimator
Args:
propensity_learner: a classifier model with probability estimation method
`predict_proba` like a sklearn LogisticRegression
learner: generic outcome regression model for both outcomes
learner_c: specific control outcome regression model
learner_t: specific treatment outcome regression model
"""
self.propensity_learner = propensity_learner
if learner is None:
if learner_c is None and learner_t is None:
self.learner_c = LassoLars()
self.learner_t = LassoLars()
else:
self.learner_c = learner_c
self.learner_t = learner_t
else:
self.learner_c = copy.deepcopy(learner)
self.learner_t = copy.deepcopy(learner)
self.delta = delta
def LARS_EN(Y, X, reg_param, reg_param1):
'''
function takes
- Y: p x 1 target variable
- X: n x p dataset
- reg_param: regularization parameter for l2-norm
- reg_param1: regularization parameter for l1-norm
function returns
- beta: 1 x p vector with coefficients
'''
# Find the number of features
p = X.shape[1]
# Create the artificial dataset for the naïve elastic net
X = np.power(1 + reg_param, -0.5) * np.vstack(
(X, np.sqrt(reg_param) * np.identity(p)))
Y = np.vstack((Y, np.zeros(shape=(p, 1))))
gamma = reg_param1 / np.sqrt(1 + reg_param)
# Center X
X = StandardScaler(with_std=False).fit_transform(X)
# Use the LARS (Efron 2004) algorithm to solve this lasso regression
lasso = LassoLars(alpha=gamma, fit_intercept=False, max_iter=1000)
lasso.fit(X, Y)
# Transform the found coefficients in the elastic net coefficients
beta = lasso.coef_ / np.sqrt(1 + reg_param)
return beta
def RunLARSScikit(q):
totalTimer = Timer()
# Load input dataset.
Log.Info("Loading dataset", self.verbose)
inputData = np.genfromtxt(self.dataset[0], delimiter=',')
responsesData = np.genfromtxt(self.dataset[1], delimiter=',')
try:
with totalTimer:
# Get all the parameters.
lambda1 = re.search("-l (\d+)", options)
lambda1 = 0.0 if not lambda1 else int(lambda1.group(1))
# Perform LARS.
model = LassoLars(alpha=lambda1)
model.fit(inputData, responsesData)
out = model.coef_
except Exception as e:
q.put(-1)
return -1
time = totalTimer.ElapsedTime()
q.put(time)
return time
def dataPreprocess():
"""
Description:使用最小角回归Lasso算法进行特征压缩
Params:
Return:
Author:
HY
Modify:
2019/6/21 16:37
"""
inputFile = 'data/data1.csv'
outputFile = 'tmp/newData.csv'
data = pd.read_csv(inputFile)
model=LassoLars(alpha=4,max_iter=1000)
model.fit(data.iloc[:,0:13],data['y'])
coefs=model.coef_
print(coefs)
# model = Lasso(alpha=1.0,max_iter=1000000,tol=0.00000001)
# model.fit(data.iloc[:, 0:13], data['y'])
# coefs=model.coef_
# print(coefs)
newColumns=[]
for index,column in enumerate(data.columns[0:13]):
if coefs[index]!=0:
newColumns.append(column)
newColumns.append(data.columns[13])
newData=pd.DataFrame(data[newColumns])#用Copy()是为了避免出现链式问题
newData['year']=list(range(1994,2014,1))
newData.to_csv(outputFile,index=False)
def train(self):
""""""
start = time.time()
print('size before truncated outliers is %d ' % len(self.TrainData))
TrainData = self.TrainData[(self.TrainData['logerror'] > self._low) & (self.TrainData['logerror'] < self._up)]
print('size after truncated outliers is %d ' % len(self.TrainData))
TrainData['longitude'] -= -118600000
TrainData['latitude'] -= 34220000
#extra_tr = pd.read_hdf(path_or_buf='%s/p21/eval_train.hdf' % self.InputDir, key='train')
#self.TrainData = pd.concat([self.TrainData, extra_tr.drop('parcelid', axis= 1)], axis = 1)
X = self.TrainData.drop(self._l_drop_cols, axis=1)
Y = self.TrainData['logerror']
self._l_train_columns = X.columns
X = X.values.astype(np.float32, copy=False)
lr = LassoLars(alpha= self._lr_alpha, max_iter= self._lr_iter, verbose= True)
self._model = lr.fit(X, Y)
end = time.time()
print('Training iterates %d, time consumed %d ' % (self._model.n_iter_, (end - start)))
self._f_eval_train_model = '{0}/{1}_{2}.pkl'.format(self.OutputDir, self.__class__.__name__,
datetime.now().strftime('%Y%m%d-%H:%M:%S'))
#with open(self._f_eval_train_model, 'wb') as o_file:
# pickle.dump(self._model, o_file, -1)
#o_file.close()
#self.TrainData = pd.concat([self.TrainData, self.ValidData[self.TrainData.columns]],
# ignore_index=True) ## ignore_index will reset the index or index will be overlaped
return
def lasso_subproblem(self, Xt):
'''
function which performs:
- 4: Sparse coding with LARS
INPUTS:
- self
- Xt, data array
- A, matrix
- B, matrix
- t, iter number
OUTPUT:
- coef
'''
print "inside lasso"
# 4: Sparse coding with LARS
from sklearn.linear_model import LassoLars
lars = LassoLars(alpha=self.alpha, verbose=False)
# self.components = np.matrix([[8,2,3,4],[1,6,1,99]])
# Xt = np.matrix([[3,1],[6,7]])
# Xt[1,1] = 9999
lars.fit(self.components, Xt)
coef = lars.coef_
# print coef
coef = (np.asmatrix(coef)).T
# Dimension control
if self.verbose > 20:
print "coef shape :", coef.shape
return coef
def OnceTest(dataMat, labelMat):
clf1 = LinearRegression()
clf1.fit(dataMat[0:99], labelMat[0:99])
labelTest1 = clf1.predict(dataMat[100:199])
print('default LinearRegression',
((labelTest1 - labelMat[100:199])**2).sum())
clf2 = Ridge(alpha=1, max_iter=100, tol=0.001)
clf2.fit(dataMat[0:99], labelMat[0:99])
labelTest2 = clf2.predict(dataMat[100:199])
print('Ridge alhpa=1 max_iter=100 tol=0.001',
((labelTest2 - labelMat[100:199])**2).sum())
clf3 = Lasso(alpha=1, max_iter=100, tol=0.001)
clf3.fit(dataMat[0:99], labelMat[0:99])
labelTest3 = clf3.predict(dataMat[100:199])
print('Lasso alhpa=1 max_iter=100 tol=0.001',
((labelTest3 - labelMat[100:199])**2).sum())
clf4 = ElasticNet(alpha=1, l1_ratio=0.5, max_iter=100, tol=1e-4)
clf4.fit(dataMat[0:99], labelMat[0:99])
labelTest4 = clf4.predict(dataMat[100:199])
print('ElasticNet alhpa=1 max_iter=100 tol=0.001',
((labelTest4 - labelMat[100:199])**2).sum())
clf5 = LassoLars(alpha=1, max_iter=100)
clf5.fit(dataMat[0:99], labelMat[0:99])
labelTest5 = clf4.predict(dataMat[100:199])
print('LassoLars alhpa=1 max_iter=100',
((labelTest5 - labelMat[100:199])**2).sum())
def LassoRegression(X_train, X_test, y_train, y_test):
regr = LassoLars(alpha=0.1)
print len(X_train.values.tolist()[0])
print len(X_train.values.tolist())
regr.fit(X_train.values.tolist(), y_train.values.tolist())
predictions = regr.predict(X_test)
return predictions
class LassoLarsPrim(primitive):
def __init__(self, random_state=0):
super(LassoLarsPrim, self).__init__(name='LassoLars')
self.hyperparams = []
self.type = 'Regressor'
self.description = "LassoLars is a lasso model implemented using the LARS algorithm, and unlike the implementation based on coordinate descent, this yields the exact solution, which is piecewise linear as a function of the norm of its coefficients."
self.hyperparams_run = {'default': True}
self.random_state = random_state
self.model = LassoLars(alpha=0.1)
self.accept_type = 'c_r'
def can_accept(self, data):
return self.can_accept_c(data, 'Regression')
def is_needed(self, data):
# data = handle_data(data)
return True
def fit(self, data):
data = handle_data(data)
self.model.fit(data['X'], data['Y'])
def produce(self, data):
output = handle_data(data)
output['predictions'] = self.model.predict(output['X'])
output['X'] = pd.DataFrame(output['predictions'], columns=[self.name+"Pred"])
final_output = {0: output}
return final_output
def trainAlgo(self):
self.model = LassoLars(alpha=self.param['alpha'],
normalize=self.param['normalize'],
fit_intercept=self.param['fit_intercept'],
max_iter=self.param['max_iter'],
positive=self.param['positive'])
self.model.fit(self.inputData['X'], self.outputData['Y'])
def RunLARSScikit(q):
totalTimer = Timer()
# Load input dataset.
Log.Info("Loading dataset", self.verbose)
inputData = np.genfromtxt(self.dataset[0], delimiter=',')
responsesData = np.genfromtxt(self.dataset[1], delimiter=',')
lambda1 = re.search("-l (\d+)", options)
lambda1 = 1.0 if not lambda1 else float(lambda1.group(1))
max_iter1 = re.search("--max_iter (\d+)", options)
max_iter1 = 500 if not max_iter1 else int(max_iter1.group(1))
eps1 = re.search("--eps (\d+)", options)
eps1 = np.finfo(float).eps if not eps1 else float(eps1.group(1))
try:
with totalTimer:
# Perform LARS.
model = LassoLars(alpha=lambda1,
max_iter=max_iter1,
eps=eps1)
model.fit(inputData, responsesData)
out = model.coef_
except Exception as e:
q.put(-1)
return -1
time = totalTimer.ElapsedTime()
q.put(time)
return time
def Lasso(x_train, y_train, x_test, y_test):
estimator = LassoLars()
estimator.fit(x_train, y_train)
y_pred = estimator.predict(x_test)
mse_score = mse(y_test, y_pred)
print("mse_score: " + str(mse_score))
r2_score = r2(y_test, y_pred)
print("r2_score: " + str(r2_score))
def predict_LarsLasso(X, y, train, test, alpha=0.1):
# Fit
lars = LassoLars(alpha)
lars.fit(X.iloc[train], y.iloc[train])
# Predict
prediction = lars.predict(X.iloc[test])
return prediction
def update_spatial_perpx(y, alpha, sub, C):
res = np.zeros_like(sub, dtype=y.dtype)
if np.sum(sub) > 0:
C = C[:, sub]
clf = LassoLars(alpha=alpha, positive=True)
coef = clf.fit(C, y).coef_
res[np.where(sub)[0]] = coef
return res
def select(self,X,y,weight,alpha=0.01):
lars = LassoLars(normalize=False,alpha=alpha)
lars.fit(X,y)
path_idx = np.argwhere((lars.coef_path_ != 0).sum(axis=0) <= self.n_features)[-1,0]
coef = lars.coef_path_[:,path_idx]
f_indices = np.argwhere(coef != 0).T[0]
if len(f_indices) == 0:
f_indices = self.select(X,y,alpha=alpha * 0.01)
return f_indices
def lasso_lars(X, y):
#train model
lars = LassoLars(alpha=0.1)\
.fit(X, y)
lars_pred = lars.predict(X)
lars_rmse = sqrt(mean_squared_error(y, lars_pred))
return lars_rmse
def __init__(self, random_state=0):
super(LassoLarsPrim, self).__init__(name='LassoLars')
self.hyperparams = []
self.type = 'Regressor'
self.description = "LassoLars is a lasso model implemented using the LARS algorithm, and unlike the implementation based on coordinate descent, this yields the exact solution, which is piecewise linear as a function of the norm of its coefficients."
self.hyperparams_run = {'default': True}
self.random_state = random_state
self.model = LassoLars(alpha=0.1)
self.accept_type = 'c_r'
def LassoLarsTest(dataMat, labelMat):
clf1 = LassoLars(alpha=1, max_iter=100)
clf1.fit(dataMat[0:99], labelMat[0:99])
labelTest1 = clf1.predict(dataMat[100:199])
print('LassoLars ', ((labelTest1 - labelMat[100:199])**2).sum())
clf2 = LassoLarsCV(max_n_alphas=10, max_iter=100)
clf2.fit(dataMat[0:99], labelMat[0:99])
labelTest2 = clf2.predict(dataMat[100:199])
print('LassoLarsCV', ((labelTest2 - labelMat[100:199])**2).sum())
def Lars_Lasso(kf,data,label,k): val=0 for train, test in kf: X_train, X_test, y_train, y_test = data[train,:], data[test,:], label[train], label[test] log = LassoLars(alpha=.1) logit = log.fit(X_train,y_train) y_pred = logit.predict(X_test) val+= metrics.mean_squared_error(y_test, y_pred) return val/3
def ll_validate_test(X, y, X_vt, y_vt):
#train model
lars = LassoLars(alpha=0.1)\
.fit(X, y)
#validate model
lars_pred_v = lars.predict(X_vt)
lars_rmse_v = sqrt(mean_squared_error(y_vt, lars_pred_v))
return lars_rmse_v
def _load_model(cls, fh):
params = _parse_literal(fh)
active = _parse_literal(fh)
coef_shape = _parse_literal(fh)
m = LassoLars().set_params(**params)
m.intercept_ = 0.0
n = int(np.prod(coef_shape)) * 8
m.coef_ = np.fromstring(fh.read(n)).reshape(coef_shape)
m.active_ = active
return m
def scaledlasso(self, X, y, intercept, lam0=None, sigma=None):
n, p = X.shape
if lam0 == None:
if p > pow(10, 6):
lam0 = 'univ'
else:
lam0 = 'quantile'
if lam0 == 'univ' or lam0 == 'universal':
lam0 = np.sqrt(2 * np.log10(p) / n)
if lam0 == 'quantile':
L = 0.1
Lold = 0
while (np.abs(L - Lold) > 0.001):
k = (L**4 + 2 * L**2)
Lold = L
L = -norm.ppf(np.min(k/p,0.99))
L = (L + Lold) / 2
if (p == 1):
L = 0.5
lam0 = np.sqrt(2 / n) * L
sigmaint = 0.1
sigmanew = 5
flag = 0
objlasso = LassoLars(fit_intercept=False,eps=0.001,fit_path=True)
objlasso.fit(X,y)
while abs(sigmaint - sigmanew) > 0.0001 and flag <= 100:
flag = flag + 1
sigmaint = np.copy(sigmanew)
lam = lam0 * sigmaint
s = lam * n
lams = objlasso.alphas_
s[np.where(s>np.max(lams))[0]]=np.max(lams)
s[np.where(s<0)[0]]=0
sfrac = (s-s[0])/(s[p-1]-s[0])
s = (s-s[0])/(s[p-1]-s[0])
hbeta = objlasso.coef_
hy = np.dot(X,hbeta)
sigmanew = np.sqrt(np.mean(np.square(y - hy)))
sigmahat = sigmanew
hlam = lam
if sigma == None:
sigmahat = np.sqrt(np.sum(np.square(y - hy)) / (n - np.sum(hbeta != 0)))
return hbeta, sigmahat
class in_lassoLars(regression):
def trainAlgo(self):
self.model = LassoLars(alpha=self.param['alpha'],
normalize=self.param['normalize'],
fit_intercept=self.param['fit_intercept'],
max_iter=self.param['max_iter'],
positive=self.param['positive'])
self.model.fit(self.inputData['X'], self.outputData['Y'])
def predictAlgo(self):
self.result['Y'] = self.model.predict(self.inputData['X'])
def metric(self):
totalTimer = Timer()
with totalTimer:
model = LassoLars(**self.build_opts)
model.fit(self.data[0], self.data[1])
out = model.coef_
metric = {}
metric["runtime"] = totalTimer.ElapsedTime()
return metric
def adaptiveLasso():
'''
Adaptive-Lasso变量选择模型
:return:
'''
inputfile = 'data/data1.csv'
data = pd.read_csv(inputfile)
# 导入AdaptiveLasso算法，要在较新的Scikit-Learn才有
from sklearn.linear_model import LassoLars
model = LassoLars()
model.fit(data.iloc[:, 0:13], data['y'])
print(model.coef_)
def fit_model_11(self,toWrite=False):
model = LassoLars(alpha=1,max_iter=5000)
for data in self.cv_data:
X_train, X_test, Y_train, Y_test = data
model.fit(X_train,Y_train)
pred = model.predict(X_test)
print("Model 11 score %f" % (logloss(Y_test,pred),))
if toWrite:
f2 = open('model11/model.pkl','w')
pickle.dump(model,f2)
f2.close()
def lasso_subproblem(self, Xt, comp):
print "inside lasso"
# 4: Sparse coding with LARS
lars = LassoLars(alpha=self.alpha, verbose=False)
lars.fit(comp, Xt)
coef = lars.coef_
# print coef
coef = (np.asmatrix(coef)).T
# Dimension control
if self.verbose > 20:
print "coef shape :", coef.shape
return coef
def lasso_sklearn(dict, target, gamma):
"""
Computes Lasso optimization
:param dict: dictionnary
:type dict: np.array
:param target: image
:type target: np.array
:param gamma: regularization factor
:type gamma: float
:rtype: np.array
"""
num_samples = target.shape[1]
patch_size = dict.shape[0]
dic_size = dict.shape[1]
gamma /= num_samples
ll = LassoLars(alpha=gamma, fit_intercept=False, normalize=False, fit_path=False)
ll.fit(dict, target)
alpha = ll.coef_
alpha = alpha.reshape(dic_size, num_samples)
return alpha
# LassoLars Regression import numpy as np from sklearn import datasets from sklearn.linear_model import LassoLars # load the iris datasets dataset = datasets.load_diabetes() # fit a LASSO using LARS model to the data model = LassoLars(alpha=0.1) model.fit(dataset.data, dataset.target) print(model) # make predictions expected = dataset.target predicted = model.predict(dataset.data) # summarize the fit of the model mse = np.mean((predicted-expected)**2) print(mse) print(model.score(dataset.data, dataset.target))
def ProcessData(df,vect1,vect2,builder):
descriptionmatrix = vect1.transform([str(x) for x in df['titledescription'].values])
locationmatrix = vect2.transform([str(x) for x in df['locationfull'].values])
# x = build_design_matrices([builder], df, return_type='dataframe', NA_action=NAAction(on_NA='drop', NA_types=[]))
y = df['SalaryNormalized'].values
#x_combo = np.hstack([np.asarray(x[0]),descriptionmatrix.toarray(),locationmatrix.toarray()])
x_combo = np.hstack([descriptionmatrix.toarray(),locationmatrix.toarray()])
return (np.asarray(y), sparse.coo_matrix(x_combo))
train = PreProcess(pd.read_csv('train.csv'))
(vect1,vect2,builder) = InitializeTransformers(train)
(y, x) = ProcessData(train, vect1, vect2,builder)
(y_test, x_test) = ProcessData(PreProcess(pd.read_csv('solution.csv')),vect1,vect2,builder)
lasso = Lasso()
lasso.fit(x,y)
y_pred = lasso.predict(x_test)
lassolars = LassoLars(alpha=2)
lassolars.fit(x.toarray(),y)
lars_pred = lassolars.predict(x_test)
print np.sqrt(mean_squared_error(y_test, y_pred))
print r2_score(y_test,y_pred)
print np.sqrt(mean_squared_error(y_test,lars_pred))
print r2_score(y_test,lars_pred)
