def main():
mean = torch.tensor(np.ones(16), dtype=torch.float32)
diag = 2 * torch.tensor(np.ones(16), dtype=torch.float32)
population = Gaussian_Distribution(mean=mean,
diag=diag,
sub=0.3,
type='chain',
slash=1)
truth = population.invcov.numpy()
n = 1000
d = population.dim
print(truth)
dist, sample, _, S = population.generate(n, numpy_like=True)
# print(S)
# print(np.array(sample))
R = np.linalg.inv(S)
# print(R)
# print(sample)
np.random.seed(0)
for alpha in np.arange(0.001, 0.010, 0.001):
model = LOCOV(alpha=alpha)
model.fit(S)
print(model.alpha)
heatmap(model.prec)
score = dict()
score['log_lik'] = []
score['AIC'] = []
score['non_zero'] = []
alpha_list = np.hstack((np.arange(0, 0.1,
0.001), np.arange(0.11, 0.3, 0.01)))
data = np.array(sample)
for alpha in alpha_list:
out_dict = cross_val_score_LOCOV(data, alpha=alpha)
score['log_lik'].append(out_dict['log_lik'])
score['AIC'].append(out_dict['AIC'])
score['non_zero'].append(out_dict['non_zero'])
plt.plot(alpha_list, score['log_lik'])
plt.show()
plt.plot(alpha_list, score['AIC'])
plt.show()
plt.plot(alpha_list, score['non_zero'])
plt.show()
def main():
mean = torch.tensor(np.ones(16), dtype=torch.float32)
diag = 2 * torch.tensor(np.ones(16), dtype=torch.float32)
population = Gaussian_Distribution(mean=mean, diag=diag, sub=0.3, type='chain', slash=1)
truth = population.invcov.numpy()
n = 1000
d = population.dim
data = pd.read_csv("samples.csv")
sample = data.iloc[1:, 1:].values
emp_cov = sample_cov(sample)
#dist, sample, _, emp_cov = population.generate(n, numpy_like=True)
model = GD()
model.fit(emp_cov, alpha=0.05, lr=0.01)
heatmap(emp_cov)
heatmap(model.prec)
def main():
mean = torch.tensor(np.ones(16), dtype=torch.float32)
diag = 2 * torch.tensor(np.ones(16), dtype=torch.float32)
population = Gaussian_Distribution(mean=mean,
diag=diag,
sub=0.3,
type='chain',
slash=1,
cross=7)
truth = population.invcov.numpy()
n = 1000
d = population.dim
print(truth)
dist, sample, _, emp_cov = population.generate(n, numpy_like=True)
print(emp_cov)
for alpha in np.arange(1e-3, 6e-3, 5e-4):
model = MarjanovicCoordinateDescent(alpha=alpha)
model.fit(emp_cov)
print(alpha)
heatmap(model.prec)
def main():
mean = torch.tensor(np.ones(16), dtype=torch.float32)
diag = torch.tensor(np.ones(16), dtype=torch.float32)
population = Gaussian_Distribution(mean=mean, diag=diag, sub=-0.3, type='chain', slash=1)
truth = population.invcov.numpy()
n = 1000
d = population.dim
print(truth)
dist, sample, _, emp_cov = population.generate(n, numpy_like=True)
model = Glasso(alpha=0.06)
model.fit(emp_cov)
heatmap(model.prec)
#heatmap(np.linalg.inv(model.prec))
#return
score = dict()
score['log_lik'] = []
score['AIC'] = []
score['non_zero'] = []
alpha_list = np.hstack((np.arange(1e-5, 0.1, 0.005), np.arange(0.1, 0.3, 0.01)))
data = np.array(sample)
for alpha in alpha_list:
out_dict = cross_val_score_My_Glasso(data, alpha=alpha)
score['log_lik'].append(out_dict['log_lik'])
score['AIC'].append(out_dict['AIC'])
score['non_zero'].append(out_dict['non_zero'])
plt.plot(alpha_list, score['log_lik'])
plt.show()
plt.plot(alpha_list, score['AIC'])
plt.show()
plt.plot(alpha_list, score['non_zero'])
plt.show()
def report(type_report):
"""[summary]
Get report according user selection
Args:
integer: [integer to be used to select chosen dataset: 1, 2, 3 or 4].
Returns:
[plot]: [plot with selected report]
"""
# Condition to ensure user selects proper petrol station id
if petrol_station_id not in get_id_by_entityType1()[0]:
result = print("Please, introduce correct Petrol Station Id!")
# Condition to ensure user selects proper petrol product
elif petrol_product_type not in product_list:
result = print("Please, introduce correct Petrol product!")
# Condition to ensure user selects proper type of report
elif select_report not in type_options:
result = print("Please, introduce correct type of report!")
else:
if (type_report == "1"):
#Report 1: Box Plot for one selected Id
result = box_price_one_id(csv_file, petrol_station_id,
petrol_product_type)
elif (type_report == "2"):
# Report 2: Box Plot for all Id
result = box_price_all_id(csv_file, petrol_product_type)
elif (type_report == "3"):
# Report 3: Line Plot showing price evolution for one selected Id
result = evolucion_precio_gasolinera(csv_file, petrol_station_id,
petrol_product_type)
elif (type_report == "4"):
# Report 4: Geographical HeatMap with the prices by product of all petrol stations
# Introduction of user's API
api = str(input("Enter your API for google maps: "))
result = heatmap(api, csv_file, petrol_product_type)
return result
def main():
mean = torch.tensor(np.ones(32), dtype=torch.float32)
diag = torch.tensor(np.ones(32), dtype=torch.float32)
population = Gaussian_Distribution(mean=mean,
diag=diag,
sub=0.25,
type='chain',
slash=1)
#truth = torch.inverse(population.cov)
truth = population.invcov
#print(x.sample())
n = 1000
d = population.dim
eps = 1e-4
print(truth)
_, _, _, S = population.generate(n)
print(S)
R = torch.inverse(S)
print(R)
x = torch.ones(3, 3)
x.requires_grad = True
l = L1_penal((x))
l.backward()
print(x.grad)
NLogL = net(dim=d, diag=1.0, beta=0.05)
param_groups = []
param_groups.append({'params': [NLogL.X]})
optimizer = GD(params=param_groups, lr=0.5, weight_decay=0)
scheduler = torch.optim.lr_scheduler.MultiStepLR(
optimizer, milestones=[40, 120, 300, 600], gamma=0.1)
performance = dict()
performance['loss'] = []
performance['iteration'] = []
max_iter = 800
for i in range(1, max_iter + 1):
out = NLogL.forward(S)
optimizer.zero_grad()
out.backward()
optimizer.step()
scheduler.step()
if i % 1 == 0:
loss = torch.norm(NLogL.X.clone().detach() - truth).item()
performance['loss'].append(loss)
performance['iteration'].append(i)
if i % 100 == 0:
print("iteration={}".format(i))
print("Frobenius loss={}".format(loss))
#print("-log_likelihood={}".format(loss.item()))
#print("gradient={}".format(NLogL.X.grad))
NLogL.X.requires_grad = False
zero = torch.zeros(d, d)
Y = NLogL.X
Y_c = Y.where(abs(Y) > eps, zero)
R_c = R.where(abs(R) > eps, zero)
print("estimation={}".format(Y))
print("censored={}".format(Y_c))
print("true_precision={}".format(truth))
print(population.cov)
print("sample_cov={}".format(S))
print("direct_inverse={}".format(R))
#plt.plot(performance['iteration'], performance['loss'])
#plt.show()
font_setting = {'fontsize': 5}
sns.heatmap(torch.abs(truth).numpy(),
annot=truth.numpy(),
annot_kws=font_setting,
cmap="YlGn",
vmin=0,
vmax=0.5,
square=True)
plt.show()
sns.heatmap(torch.abs(Y_c).numpy(),
annot=Y_c.numpy(),
annot_kws=font_setting,
cmap="YlGn",
vmin=0,
vmax=0.5,
square=True)
plt.show()
sns.heatmap(torch.abs(R_c).numpy(),
annot=R_c.numpy(),
annot_kws=font_setting,
cmap="YlGn",
vmin=0,
vmax=0.5,
square=True)
plt.show()
heatmap(Y_c.numpy())