def calc_n_shapley_values(n_feats, n_samples, n_iter, data_type, cf_name, overwrite=False, data_dir="result_data_sunnies"):
"""
Returns a nested list of shapley values (per player) per iteration;
[[v1... vn], [v1...vn], [v1...vn], ...]
I.e. the length of the list is equal to n_iter
"""
players = list(range(n_feats))
filename = f"{data_dir}/{n_feats}_feats_{n_samples}_samples_{n_iter}_iter_{cf_name}.npy"
if not overwrite and os.path.exists(filename):
return numpy.load(filename)
all_shaps = []
for _i in range(n_iter):
x, y = data.make_data(n_feats, n_samples, data_type)
_shapley_values = calc_shapley_values(x, y, cf_name)
all_shaps.append(_shapley_values)
numpy.save(filename, all_shaps)
return all_shaps
my_array = {}
my_url = input("Ссылка которую нужно сократить: ")
my_new_name = input("Новое имя ссылки: ")
emulator.adding_to_dict(my_new_name, my_url, my_array)
while True:
new_link_request = input("Хотите добавить еще ссылку?(y/n): ")
if new_link_request == "y":
my_url = input("Ссылка которую нужно сократить: ")
my_new_name = input("Новое имя ссылки: ")
emulator.adding_to_dict(my_new_name, my_url, my_array)
elif new_link_request == "n":
break
data.make_data(my_new_name, my_url)
while True:
link_request = input("Хотите получить свою ссылку(y/n): ")
if link_request == "y":
name = input("Имя: ")
emulator.get_linc_from_dict(name, my_array)
elif link_request == "n":
break
while True:
ans = input('Показать БД?(y/n): ')
if ans == 'y':
with shelve.open('data') as db:
data = dict(db.items())
print(data)
# Number of irrelevant features
N_IRRELEVANT = 0
# Models to use
MODELS = [Ridge, KNeighborsRegressor]
# Model complexity
COMPLEXITIES = [1.0, 5]
########
# Main #
########
if __name__ == '__main__':
# Generate data
X, y = make_data(N_DATASETS * N_SAMPLES, N_IRRELEVANT)
# Plot data
scatter_plot(X[:, 0], y, 'Q3d_data')
# Calculate expected error and its terms for
# each model
for model, complexity in zip(MODELS, COMPLEXITIES):
# Create the protocol
p = Protocol(X, y)
# Train models
p.train(model, complexity, N_DATASETS)
# Get error and its terms
noise, s_bias, var, exp_error = p.eval()
from data import make_data
import plotly
import plotly.graph_objs as go
file = open('FRvideos.csv', 'r', encoding="utf8")
dataset = make_data(file)
print(dataset)
#вивести стовпчикову діаграму з кілкістю відео за кожну дату
new_dict = dict()
for category_id in dataset:
for date in dataset[category_id]:
if date in new_dict:
new_dict[date] += len(dataset[category_id][date].values())
else:
new_dict[date] = len(dataset[category_id][date].values())
diagram = go.Bar(x=list(new_dict.keys()), y=list(new_dict.values()))
fig = go.Figure(data=[diagram])
plotly.offline.plot(fig, filename='graph1.html')
#вивести кругову діаграму з кількістю відео в кожній категорії
new_new_dict = dict()
for category_id in dataset:
for date in dataset[category_id]:
if category_id in new_new_dict:
new_new_dict[category_id] += len(
dataset[category_id][date].values())
else:
new_new_dict[category_id] = len(
dataset[category_id][date].values())
stack.extend([
ZeroCenter(),
LinearSymplecticTwoByTwo(),
SymplecticAdditiveCoupling(shift_model=IrrotationalMLP())
])
#SymplecticAdditiveCoupling(shift_model=MLP())])
T = Chain(stack)
step = tf.get_variable("global_step", [],
tf.int64,
tf.zeros_initializer(),
trainable=False)
with tf.Session() as sess:
z = make_data(settings, sess)
loss = make_loss(settings, T, z)
train_op = make_train_op(settings, loss, step)
# sess.run(tf.global_variables_initializer())
# Set the ZeroCenter bijectors to training mode:
for i, bijector in enumerate(T.bijectors):
if hasattr(bijector, 'is_training'):
T.bijectors[i].is_training = True
tf.contrib.training.train(train_op,
logdir=settings['log_dir'],
save_checkpoint_secs=60)
import matplotlib.pyplot as plt
from matplotlib.colors import ListedColormap
from sklearn import neighbors, datasets
from data import make_data
from plot import plot
# (X^-1 * X.T) * y
# X * X.T
# [
# [1, 3],
# [2, 4]
# ... ]
# β=(X.T*X)^−1 * X.T * y
X_train, y_train, X_test, y_test = make_data(n_samples=6)
intercept = np.ones(shape=y_train.shape).reshape(-1, 1)
print("intercept:", intercept)
X_train = np.concatenate((X_train, intercept), 1)
print("\nX_train\n", X_train)
print(X_train.shape)
a = X_train.T
print("\nX_train.T\n", a)
print(a.shape)
a = a.dot(X_train)
print("\nX_train.T.dot(X_train)\n", a)
print(a.shape)
a = inv(a)
print("\nX_train.T.dot(X_train)*-1\n", a)
def get_dataset(sample_number):
X, y = make_data(sample_number, random_state=get_random_state())
return X, y
print('Cost : ', cost(C, X, Z))
if __name__ == "__main__":
# Loading the existing data
if real_data:
temp_X, temp_Y = load_file(load_data)
random.shuffle(temp_X)
random.shuffle(temp_Y)
U, y = removeDups(temp_X, temp_Y)
# Synthetic Data
else:
U, y, C_, Z_, ids_ = make_data(5, 0, 8, 50)
# # X_train, X_test, y_train, y_test = train_test_split(np.array(temp_X), np.array(temp_Y), test_size=0.33, random_state=42)
# # print(X_test.shape)
# # data is finally in U and labels in y
# print('u shape ', len(U),',',len(U[0]))
# print(U[0][0])
# print(U[1][0])
# print(U[2][0])
# # print(LS(U, [U[0]], 1)[0])
# # print(cost_km([U[1]], U))
if LSAlgo in RunAlgos:
Z = [U[x[0]] for x in dists[-z:]]
X = [x[0] for x in dists[:-z]] # storing index of point in U
cNum = [0 for _ in range(k)]
C = np.zeros((k, len(U[0])))
for i in X:
cNum[cIds[i]] += 1 # update no of points in cluster
C[cIds[i]] = C[cIds[i]] + U[i]
for j in range(k):
if cNum[j] != 0:
C[j] = C[j] / cNum[j]
else:
print('empty')
return C, Z
if __name__ == "__main__":
random.seed(0)
np.random.seed(0)
U, y, C_, Z_, ids_ = make_data(5, 10, 10, 100, num_points=10)
C, Z = kmeans_minus(U, 3, 5)
plotGraph(U, C, Z, "./Plots/KMeans_")
'''
4 -> make_data(5, 0, 10, 50)
3 -> make_data(5, 10, 10, 50)
2 -> make_data(5, 0, 8, 50)
1 -> make_data(5, 0, 10, 100)
'''