def initialize_diary():
diary = Diary(name='digits_vs_letters',
path='results',
overwrite=False,
fig_format='svg')
diary.add_notebook('training', verbose=True)
diary.add_notebook('validation', verbose=True)
return diary
def setUp(self):
self.EPSILON = 0.01
sub1 = Subject("Biologia",
[[3.5, 2.5, 5.0], [1, 1, 1, 1, 1, 1, 1, 1, 0]])
sub2 = Subject("Matematyka",
[[2.5, 3.0, 2.5], [1, 1, 0, 1, 0, 1, 0, 1, 1]])
self.subjects = [sub1, sub2]
self.andrzej = Student("Andrzej Abacki", self.subjects)
self.diary = Diary("AGH", 2016, "data.json")
def __init__(self):
Gtk.Window.__init__(self, title='Diary')
self.set_default_size(800, 600)
self.connect('destroy', Gtk.main_quit)
main_box = Gtk.Box()
size_group = Gtk.SizeGroup(Gtk.SizeGroupMode.HORIZONTAL)
# initializing all relevant classes
diary = Diary("Test User")
textview = Textview()
header = Header()
searchbar = Searchbar()
sidebox = Sidebox()
# transfer classes to linker
linker = Linker(diary, header, sidebox, searchbar, textview)
# transfer linker to header and sidebox
header.set_connection_linker(linker)
sidebox.set_connection_linker(linker)
sidebox.update_year()
searchbar.set_connection_linker(linker)
searchbar.set_revealer_signal()
# connect size_group to header and sidebox
header.set_size_group(size_group)
sidebox.set_size_group(size_group)
sidebox.set_revealer_signal()
# setup relevant buttons in header
header.set_backbutton()
header.set_forwardbutton()
header.set_searchbutton()
header.set_addbutton()
header.set_editbutton()
self.set_titlebar(header)
# create new side_box to add searchbar and sidebox
# and add it to beginning of mainbox
side_box = Gtk.VBox()
side_box.pack_start(searchbar, False, False, 0)
side_box.pack_start(sidebox, True, True, 0)
side_box.set_hexpand(False)
main_box.pack_start(side_box, False, False, 0)
# add separator between side_box and textview
separator = Gtk.HSeparator()
separator.set_size_request(1, 0)
main_box.pack_start(separator, False, False, 0)
# add textview to end of mainbox
main_box.pack_start(textview, False, True, 0)
self.add(main_box)
def initialize_diary():
diary = Diary(STUDENTS, CLASSES, DATES)
for i in xrange(NUMBER_OF_SCORES):
diary.add_score(random.choice(CLASSES), random.choice(STUDENTS),
random.choice(SCORES))
for clazz in CLASSES:
for date in DATES:
for student in STUDENTS:
if random.choice([True, False]):
diary.add_attendance(clazz, student, date)
return diary
def do_it(self, ev):
login = self.login_edit.get()
password = self.password_edit.get()
diary_id = self.diary_id_edit.get()
filename = self.filename_edit.get()
split_type = self.split_type.get()
if not login:
messagebox.showinfo("Error", "Логин не задан")
return
if not diary_id:
messagebox.showinfo("Error", "Адрес сообщества не задан")
return
if not filename:
messagebox.showinfo("Error", "Путь к файлу не задан")
return
api = Diary()
try:
api.login(login, password)
text_with_header = util.load(filename)
prefix = os.path.splitext(filename)[0]
text_with_header = util.fix_characters(text_with_header)
header, text = find_header(text_with_header)
if split_type == 1:
post, comments = split_text_with_comments(header, text)
util.store(prefix + "_post.txt", post)
for i, comment in enumerate(comments):
util.store(prefix + "_comment_%d.txt" % (i+1), comment)
# Send to diary
post_id = api.new_post(post, diary_id)
for comment in comments:
api.add_comment(post_id, comment)
if len(comments) > 0:
messagebox.showinfo("Info", "Пост успешно опубликован, тексты комментариев ищите в файлах *comment_N.txt")
else:
messagebox.showinfo("Info", "Пост успешно опубликован")
else:
posts = split_text_with_posts(header, text)
for i, post in enumerate(posts):
util.store(prefix + "_post_%d.txt" % (i + 1), post)
# Send to diary
for post in posts:
api.new_post(post, diary_id)
messagebox.showinfo("Info", "Посты успешно опубликованы. Тексты продублированы в файлы *post_N.txt")
except Exception as e:
messagebox.showinfo("Error", str(e))
return
def _retrieve_from_diary(year, number):
"""Retrieve the data from the data indicated.
"""
url_base = compose_url(URL_BASE, year, number)
dia = Diary(year, number, url_base)
print "Retrieving contents from: %s" % url_base
for section in sorted(SECTIONS.keys()):
url = compose_url(url_base, section)
dia.add_section(SECTIONS[section])
WScrap.scrap_page(url, dia)
return dia
def main(dataset_names=None):
if dataset_names is None:
dataset_names = [
'autos',
'car',
'cleveland',
'dermatology',
'ecoli',
'flare',
'glass',
'led7digit',
'lymphography',
'nursery',
'page-blocks',
'pendigits',
'satimage',
'segment',
#'shuttle',
'vehicle',
'vowel',
'yeast',
'zoo',
'auslan'
]
seed_num = 42
mc_iterations = 5
n_folds = 2
estimator_type = "svm"
# Diary to save the partial and final results
diary = Diary(name='results_Krawczyk2015',
path='results',
overwrite=False,
fig_format='svg')
# Hyperparameters for this experiment (folds, iterations, seed)
diary.add_notebook('parameters', verbose=True)
# Summary for each dataset
diary.add_notebook('datasets', verbose=False)
# Partial results for validation
diary.add_notebook('validation', verbose=True)
# Final results
diary.add_notebook('summary', verbose=True)
columns = ['dataset', 'method', 'mc', 'test_fold', 'acc']
df = MyDataFrame(columns=columns)
diary.add_entry('parameters', [
'seed', seed_num, 'mc_it', mc_iterations, 'n_folds', n_folds,
'estimator_type', estimator_type
])
data = Data(dataset_names=dataset_names)
for i, (name, dataset) in enumerate(data.datasets.iteritems()):
np.random.seed(seed_num)
dataset.print_summary()
diary.add_entry('datasets', [dataset.__str__()])
accuracies = np.zeros(mc_iterations * n_folds)
for mc in np.arange(mc_iterations):
skf = StratifiedKFold(dataset.target,
n_folds=n_folds,
shuffle=True)
test_folds = skf.test_folds
for test_fold in np.arange(n_folds):
x_train, y_train, x_test, y_test = separate_sets(
dataset.data, dataset.target, test_fold, test_folds)
if estimator_type == "svm":
est = OneClassSVM(nu=0.5, gamma=0.5)
elif estimator_type == "gmm":
est = GMM(n_components=3)
bc = BackgroundCheck(estimator=est)
oc = OcDecomposition(base_estimator=bc)
oc.fit(x_train, y_train)
accuracy = oc.accuracy(x_test, y_test)
accuracies[mc * n_folds + test_fold] = accuracy
diary.add_entry('validation', [
'dataset', name, 'method', 'our', 'mc', mc, 'test_fold',
test_fold, 'acc', accuracy
])
df = df.append_rows([[name, 'our', mc, test_fold, accuracy]])
df = df.convert_objects(convert_numeric=True)
table = df.pivot_table(values=['acc'],
index=['dataset'],
columns=['method'],
aggfunc=[np.mean, np.std])
diary.add_entry('summary', [table])
def main():
dataset_names = ['diabetes', 'ecoli', 'glass', 'heart-statlog',
'ionosphere', 'iris', 'letter', 'mfeat-karhunen',
'mfeat-morphological', 'mfeat-zernike', 'optdigits',
'pendigits', 'sonar', 'vehicle', 'waveform-5000']
data = Data(dataset_names=dataset_names)
diary = Diary(name='hempstalk', path='results', overwrite=False,
fig_format='svg')
diary.add_notebook('cross_validation')
# Columns for the DataFrame
columns=['Dataset', 'MC iteration', 'N-fold id', 'Actual class', 'Model',
'AUC', 'Prior']
# Create a DataFrame to record all intermediate results
df = pd.DataFrame(columns=columns)
mc_iterations = 10
n_folds = 10
gammas = {"diabetes":0.00005, "ecoli":0.1, "glass":0.005,
"heart-statlog":0.0001, "ionosphere":0.00005, "iris":0.0005,
"letter":0.000005, "mfeat-karhunen":0.0001,
"mfeat-morphological":0.0000001, "mfeat-zernike":0.000001,
"optdigits":0.00005, "pendigits":0.000001, "sonar":0.001,
"vehicle":0.00005, "waveform-5000":0.001}
for i, (name, dataset) in enumerate(data.datasets.iteritems()):
print('Dataset number {}'.format(i))
data.sumarize_datasets(name)
for mc in np.arange(mc_iterations):
skf = StratifiedKFold(dataset.target, n_folds=n_folds, shuffle=True)
test_folds = skf.test_folds
for test_fold in np.arange(n_folds):
x_train, y_train, x_test, y_test = separate_sets(
dataset.data, dataset.target, test_fold, test_folds)
n_training = np.alen(y_train)
w_auc_fold_dens = 0
w_auc_fold_bag = 0
w_auc_fold_com = 0
prior_sum = 0
for actual_class in dataset.classes:
tr_class = x_train[y_train == actual_class, :]
tr_class_unique_values = [np.unique(tr_class[:,column]).shape[0] for column in
range(tr_class.shape[1])]
cols_keep = np.where(np.not_equal(tr_class_unique_values,1))[0]
tr_class = tr_class[:,cols_keep]
x_test_cleaned = x_test[:,cols_keep]
t_labels = (y_test == actual_class).astype(int)
prior = np.alen(tr_class) / n_training
if np.alen(tr_class) > 1 and not all(t_labels == 0):
prior_sum += prior
n_c = tr_class.shape[1]
if n_c > np.alen(tr_class):
n_c = np.alen(tr_class)
# Train a Density estimator
model_gmm = GMM(n_components=1, covariance_type='diag')
model_gmm.fit(tr_class)
sv = OneClassSVM(nu=0.1, gamma=0.5)
bc = BackgroundCheck(estimator=sv)
bc.fit(tr_class)
svm_scores = bc.predict_proba(x_test_cleaned)[:, 1]
# Generate artificial data
new_data = model_gmm.sample(np.alen(tr_class))
# Train a Bag of Trees
bag = BaggingClassifier(
base_estimator=DecisionTreeClassifier(),
n_estimators=10)
new_data = np.vstack((tr_class, new_data))
y = np.zeros(np.alen(new_data))
y[:np.alen(tr_class)] = 1
bag.fit(new_data, y)
# Combine the results
probs = bag.predict_proba(x_test_cleaned)[:, 1]
scores = model_gmm.score(x_test_cleaned)
com_scores = (probs / np.clip(1.0 - probs, np.float32(1e-32), 1.0)) * (scores-scores.min())
# Generate our new data
# FIXME solve problem with #samples < #features
pca=True
if tr_class.shape[0] < tr_class.shape[1]:
pca=False
our_new_data = reject.create_reject_data(
tr_class, proportion=1,
method='uniform_hsphere', pca=pca,
pca_variance=0.99, pca_components=0,
hshape_cov=0, hshape_prop_in=0.99,
hshape_multiplier=1.5)
our_new_data = np.vstack((tr_class, our_new_data))
y = np.zeros(np.alen(our_new_data))
y[:np.alen(tr_class)] = 1
# Train Our Bag of Trees
our_bag = BaggingClassifier(
base_estimator=DecisionTreeClassifier(),
n_estimators=10)
our_bag.fit(our_new_data, y)
# Combine the results
our_probs = our_bag.predict_proba(x_test_cleaned)[:, 1]
our_comb_scores = (our_probs / np.clip(1.0 - our_probs,
np.float32(1e-32), 1.0)) * (scores-scores.min())
# Scores for the Density estimator
auc_dens = roc_auc_score(t_labels, scores)
# Scores for the Bag of trees
auc_bag = roc_auc_score(t_labels, probs)
# Scores for the Combined model
auc_com = roc_auc_score(t_labels, com_scores)
# Scores for our Bag of trees (trained on our data)
auc_our_bag = roc_auc_score(t_labels, our_probs)
# Scores for our Bag of trees (trained on our data)
auc_our_comb = roc_auc_score(t_labels, our_comb_scores)
# Scores for the Background Check with SVm
auc_svm = roc_auc_score(t_labels, svm_scores)
# Create a new DataFrame to append to the original one
dfaux = pd.DataFrame([[name, mc, test_fold, actual_class,
'Combined', auc_com, prior],
[name, mc, test_fold, actual_class,
'P(T$|$X)', auc_bag, prior],
[name, mc, test_fold, actual_class,
'P(X$|$A)', auc_dens, prior],
[name, mc, test_fold, actual_class,
'Our Bagg', auc_our_bag, prior],
[name, mc, test_fold, actual_class,
'Our Combined', auc_our_comb, prior],
[name, mc, test_fold, actual_class,
'SVM_BC', auc_svm, prior]],
columns=columns)
df = df.append(dfaux, ignore_index=True)
# generate_and_save_plots(t_labels, scores, diary, name, mc, test_fold,
# actual_class, 'P(X$|$A)')
# generate_and_save_plots(t_labels, probs, diary, name, mc, test_fold,
# actual_class, 'P(T$|$X)')
# generate_and_save_plots(t_labels, com_scores, diary, name, mc, test_fold,
# actual_class, 'Combined')
# generate_and_save_plots(t_labels, our_probs, diary, name, mc, test_fold,
# actual_class, 'Our_Bagg')
# generate_and_save_plots(t_labels, our_comb_scores, diary, name, mc, test_fold,
# actual_class, 'Our_Combined')
# generate_and_save_plots(t_labels, svm_scores, diary,
# name, mc, test_fold,
# actual_class, 'SVM_BC')
# Convert values to numeric
df = df.convert_objects(convert_numeric=True)
# Group everything except classes
dfgroup_classes = df.groupby(by=['Dataset', 'MC iteration', 'N-fold id',
'Model'])
# Compute the Prior sum for each dataset, iteration and fold
df['Prior_sum'] = dfgroup_classes['Prior'].transform(np.sum)
# Compute the individual weighted AUC per each class and experiment
df['wAUC'] = df.Prior * df.AUC / df.Prior_sum
# Sum the weighted AUC of each class per each experiment
series_wAUC = dfgroup_classes['wAUC'].sum()
# Transform the series to a DataFrame
df_wAUC = series_wAUC.reset_index(inplace=False)
# Compute mean and standard deviation of wAUC per Dataset and model
final_results = df_wAUC.groupby(['Dataset', 'Model'])['wAUC'].agg([np.mean,
np.std])
# Transform the series to a DataFrame
final_results.reset_index(inplace=True)
# Represent the results in a table format
final_table = final_results.pivot_table(values=['mean', 'std'],
index=['Dataset'], columns=['Model'])
# Export the results in a csv and LaTeX file
export_results(final_table)
y = np.hstack((np.ones(np.alen(x)), np.zeros(np.alen(r)))).T
model_rej.fit(xr, y)
return model_rej
def train_classifier_model(x, y):
model_clas = svm.SVC(probability=True)
#model_clas = tree.DecisionTreeClassifier(max_depth=3)
model_clas = model_clas.fit(x, y)
return model_clas
if __name__ == "__main__":
diary = Diary(name='test_rgrpg',
path='results',
overwrite=False,
fig_format='svg')
diary.add_notebook('training')
diary.add_notebook('validation')
# for i in [6]: #range(1,4):
n_iterations = 1
n_thresholds = 100
accuracies = np.empty((n_iterations, n_thresholds))
recalls = np.empty((n_iterations, n_thresholds))
for example in [2, 3, 4, 5, 6, 7, 8, 9]:
np.random.seed(42)
print('Runing example = {}'.format(example))
for iteration in range(n_iterations):
#####################################################
# - get students average score in class
# - hold students name and surname
# - Count total attendance of student
# The default interface for interaction should be python interpreter.
# Please, use your imagination and create more functionalities.
# Your project should be able to handle entire school.
# If you have enough courage and time, try storing (reading/writing)
# data in text files (YAML, JSON).
# If you have even more courage, try implementing user interface.
#
#Try to expand your implementation as best as you can.
#Think of as many features as you can, and try implementing them.
#Make intelligent use of pythons syntactic sugar (overloading, iterators, generators, etc)
#Most of all: CREATE GOOD, RELIABLE, READABLE CODE.
#The goal of this task is for you to SHOW YOUR BEST python programming skills.
#Impress everyone with your skills, show off with your code.
#
#Your program must be runnable with command "python task.py".
#Show some usecases of your library in the code (print some things)
#
#When you are done upload this code to your github repository.
#
#Delete these comments before commit!
#Good luck.
from diary import Diary, Student, SchoolClass
diary = Diary()
schoolclass = SchoolClass("biology")
student = Student("majlosz", "ef")
#schoolclass.add_students([student])
def main(dataset_names=None,
estimator_type="gmm",
mc_iterations=20,
n_folds=5,
n_ensemble=100,
seed_num=42):
if dataset_names is None:
# All the datasets used in Li2014
datasets_li2014 = [
'abalone', 'balance-scale', 'credit-approval', 'dermatology',
'ecoli', 'german', 'heart-statlog', 'hepatitis', 'horse',
'ionosphere', 'lung-cancer', 'libras-movement', 'mushroom',
'diabetes', 'landsat-satellite', 'segment', 'spambase', 'wdbc',
'wpbc', 'yeast'
]
datasets_hempstalk2008 = [
'diabetes', 'ecoli', 'glass', 'heart-statlog', 'ionosphere',
'iris', 'letter', 'mfeat-karhunen', 'mfeat-morphological',
'mfeat-zernike', 'optdigits', 'pendigits', 'sonar', 'vehicle',
'waveform-5000'
]
datasets_others = [
'diabetes', 'ecoli', 'glass', 'heart-statlog', 'ionosphere',
'iris', 'letter', 'mfeat-karhunen', 'mfeat-morphological',
'mfeat-zernike', 'optdigits', 'pendigits', 'sonar', 'vehicle',
'waveform-5000', 'scene-classification', 'tic-tac', 'autos', 'car',
'cleveland', 'dermatology', 'flare', 'page-blocks', 'segment',
'shuttle', 'vowel', 'zoo', 'abalone', 'balance-scale',
'credit-approval', 'german', 'hepatitis', 'lung-cancer'
]
# Datasets that we can add but need to be reduced
datasets_to_add = ['MNIST']
dataset_names = list(
set(datasets_li2014 + datasets_hempstalk2008 + datasets_others))
# Diary to save the partial and final results
diary = Diary(name='results_Li2014',
path='results',
overwrite=False,
fig_format='svg')
# Hyperparameters for this experiment (folds, iterations, seed)
diary.add_notebook('parameters', verbose=True)
# Summary for each dataset
diary.add_notebook('datasets', verbose=False)
# Partial results for validation
diary.add_notebook('validation', verbose=True)
# Final results
diary.add_notebook('summary', verbose=True)
columns = ['dataset', 'method', 'mc', 'test_fold', 'acc', 'logloss']
df = MyDataFrame(columns=columns)
diary.add_entry('parameters', [
'seed', seed_num, 'mc_it', mc_iterations, 'n_folds', n_folds,
'n_ensemble', n_ensemble, 'estimator_type', estimator_type
])
data = Data(dataset_names=dataset_names)
for name, dataset in data.datasets.iteritems():
if name in ['letter', 'shuttle']:
dataset.reduce_number_instances(0.1)
export_datasets_description_to_latex(data, path=diary.path)
for i, (name, dataset) in enumerate(data.datasets.iteritems()):
np.random.seed(seed_num)
dataset.print_summary()
diary.add_entry('datasets', [dataset.__str__()])
for mc in np.arange(mc_iterations):
skf = StratifiedKFold(dataset.target,
n_folds=n_folds,
shuffle=True)
test_folds = skf.test_folds
for test_fold in np.arange(n_folds):
x_train, y_train, x_test, y_test = separate_sets(
dataset.data, dataset.target, test_fold, test_folds)
# Binary discriminative classifier
sv = SVC(kernel='linear', probability=True)
# Density estimator for the background check
if estimator_type == "svm":
gamma = 1.0 / x_train.shape[1]
est = OneClassSVM(nu=0.1, gamma=gamma)
elif estimator_type == "gmm":
est = GMM(n_components=1)
elif estimator_type == "gmm3":
est = GMM(n_components=3)
elif estimator_type == "mymvn":
est = MyMultivariateNormal()
# Multiclass discriminative model with one-vs-one binary class.
ovo = OvoClassifier(base_classifier=sv)
classifier = ConfidentClassifier(classifier=ovo,
estimator=est,
mu=0.5,
m=0.5)
ensemble = Ensemble(base_classifier=classifier,
n_ensemble=n_ensemble)
# classifier = ConfidentClassifier(classifier=sv,
# estimator=est, mu=0.5,
# m=0.5)
# ovo = OvoClassifier(base_classifier=classifier)
# ensemble = Ensemble(base_classifier=ovo,
# n_ensemble=n_ensemble)
xs_bootstrap, ys_bootstrap = ensemble.fit(x_train, y_train)
accuracy = ensemble.accuracy(x_test, y_test)
log_loss = ensemble.log_loss(x_test, y_test)
diary.add_entry('validation', [
'dataset', name, 'method', 'our', 'mc', mc, 'test_fold',
test_fold, 'acc', accuracy, 'logloss', log_loss
])
df = df.append_rows(
[[name, 'our', mc, test_fold, accuracy, log_loss]])
# Li2014: EP-CC model
# The classification confidence is used in learning the weights
# of the base classifier as well as in weighted voting.
ensemble_li = Ensemble(n_ensemble=n_ensemble, lambd=1e-8)
ensemble_li.fit(x_train,
y_train,
xs=xs_bootstrap,
ys=ys_bootstrap)
accuracy_li = ensemble_li.accuracy(x_test, y_test)
log_loss_li = ensemble_li.log_loss(x_test, y_test)
diary.add_entry('validation', [
'dataset', name, 'method', 'Li2014', 'mc', mc, 'test_fold',
test_fold, 'acc', accuracy_li, 'logloss', log_loss_li
])
df = df.append_rows(
[[name, 'Li2014', mc, test_fold, accuracy_li,
log_loss_li]])
export_summary(df, diary)
def main(dataset_names=None,
estimator_type="kernel",
mc_iterations=1,
n_folds=10,
seed_num=42):
if dataset_names is None:
dataset_names = ['glass', 'hepatitis', 'ionosphere', 'vowel']
bandwidths_o_norm = {
'glass': 0.09,
'hepatitis': 0.105,
'ionosphere': 0.039,
'vowel': 0.075
}
bandwidths_bc = {
'glass': 0.09,
'hepatitis': 0.105,
'ionosphere': 0.039,
'vowel': 0.0145
}
bandwidths_t_norm = {
'glass': 0.336,
'hepatitis': 0.015,
'ionosphere': 0.0385,
'vowel': 0.0145
}
tuned_mus = {
'glass': [0.094, 0.095, 0.2, 0.0, 0.0, 0.1],
'vowel': [0.0, 0.0, 0.5, 0.5, 0.5, 0.0]
}
tuned_ms = {
'glass': [1.0, 1.0, 1.0, 1.0, 1.0, 1.0],
'vowel': [1.0, 1.0, 1.0, 1.0, 1.0, 1.0]
}
bandwidth_o_norm = 0.05
bandwidth_t_norm = 0.05
bandwidth_bc = 0.05
# Diary to save the partial and final results
diary = Diary(name='results_Tax2008',
path='results',
overwrite=False,
fig_format='svg')
# Hyperparameters for this experiment (folds, iterations, seed)
diary.add_notebook('parameters', verbose=True)
# Summary for each dataset
diary.add_notebook('datasets', verbose=False)
# Partial results for validation
diary.add_notebook('validation', verbose=True)
# Final results
diary.add_notebook('summary', verbose=True)
columns = ['dataset', 'method', 'mc', 'test_fold', 'acc']
df = MyDataFrame(columns=columns)
diary.add_entry('parameters', [
'seed', seed_num, 'mc_it', mc_iterations, 'n_folds', n_folds,
'estimator_type', estimator_type, 'bw_o', bandwidth_o_norm, 'bw_t',
bandwidth_t_norm, 'bw_bc', bandwidth_bc
])
data = Data(dataset_names=dataset_names)
for name, dataset in data.datasets.iteritems():
if name in ['letter', 'shuttle']:
dataset.reduce_number_instances(0.1)
export_datasets_description_to_latex(data, path=diary.path)
for i, (name, dataset) in enumerate(data.datasets.iteritems()):
np.random.seed(seed_num)
dataset.print_summary()
diary.add_entry('datasets', [dataset.__str__()])
# accuracies_tuned = np.zeros(mc_iterations * n_folds)
# if name in bandwidths_o_norm.keys():
# bandwidth_o_norm = bandwidths_o_norm[name]
# bandwidth_t_norm = bandwidths_t_norm[name]
# bandwidth_bc = bandwidths_bc[name]
# else:
# bandwidth_o_norm = np.mean(bandwidths_o_norm.values())
# bandwidth_t_norm = np.mean(bandwidths_t_norm.values())
# bandwidth_bc = np.mean(bandwidths_bc.values())
for mc in np.arange(mc_iterations):
skf = StratifiedKFold(dataset.target,
n_folds=n_folds,
shuffle=True)
test_folds = skf.test_folds
for test_fold in np.arange(n_folds):
x_train, y_train, x_test, y_test = separate_sets(
dataset.data, dataset.target, test_fold, test_folds)
# if name in ['glass', 'hepatitis', 'ionosphere', 'thyroid',
# 'iris', 'heart-statlog', 'diabetes', 'abalone',
# 'mushroom', 'spambase']:
x_test, y_test = generate_outliers(x_test, y_test)
# elif name == 'vowel':
# x_train = x_train[y_train <= 5]
# y_train = y_train[y_train <= 5]
# y_test[y_test > 5] = 6
# elif dataset.n_classes > 2:
# x_train = x_train[y_train <= dataset.n_classes/2]
# y_train = y_train[y_train <= dataset.n_classes/2]
# y_test[y_test > dataset.n_classes/2] = dataset.n_classes+1
# else:
# continue
if estimator_type == "svm":
est = OneClassSVM(nu=0.5, gamma=1.0 / x_train.shape[1])
elif estimator_type == "gmm":
est = GMM(n_components=1)
elif estimator_type == "gmm3":
est = GMM(n_components=3)
elif estimator_type == "kernel":
est = MyMultivariateKernelDensity(kernel='gaussian',
bandwidth=bandwidth_bc)
estimators = None
bcs = None
if estimator_type == "kernel":
estimators, bcs = fit_estimators(
MyMultivariateKernelDensity(kernel='gaussian',
bandwidth=bandwidth_bc),
x_train, y_train)
# Untuned background check
bc = BackgroundCheck(estimator=est, mu=0.0, m=1.0)
oc = OcDecomposition(base_estimator=bc)
if estimators is None:
oc.fit(x_train, y_train)
else:
oc.set_estimators(bcs, x_train, y_train)
accuracy = oc.accuracy(x_test, y_test)
diary.add_entry('validation', [
'dataset', name, 'method', 'BC', 'mc', mc, 'test_fold',
test_fold, 'acc', accuracy
])
df = df.append_rows([[name, 'BC', mc, test_fold, accuracy]])
e = MyMultivariateKernelDensity(kernel='gaussian',
bandwidth=bandwidth_o_norm)
oc_o_norm = OcDecomposition(base_estimator=e,
normalization="O-norm")
if estimators is None:
oc_o_norm.fit(x_train, y_train)
else:
oc_o_norm.set_estimators(estimators, x_train, y_train)
accuracy_o_norm = oc_o_norm.accuracy(x_test, y_test)
diary.add_entry('validation', [
'dataset', name, 'method', 'O-norm', 'mc', mc, 'test_fold',
test_fold, 'acc', accuracy_o_norm
])
df = df.append_rows(
[[name, 'O-norm', mc, test_fold, accuracy_o_norm]])
e = MyMultivariateKernelDensity(kernel='gaussian',
bandwidth=bandwidth_t_norm)
oc_t_norm = OcDecomposition(base_estimator=e,
normalization="T-norm")
if estimators is None:
oc_t_norm.fit(x_train, y_train)
else:
oc_t_norm.set_estimators(estimators, x_train, y_train)
accuracy_t_norm = oc_t_norm.accuracy(x_test, y_test)
diary.add_entry('validation', [
'dataset', name, 'method', 'T-norm', 'mc', mc, 'test_fold',
test_fold, 'acc', accuracy_t_norm
])
df = df.append_rows(
[[name, 'T-norm', mc, test_fold, accuracy_t_norm]])
# Tuned background check
# if name in tuned_mus.keys():
# mus = tuned_mus[name]
# ms = tuned_ms[name]
# else:
# mus = None
# ms = None
# bc = BackgroundCheck(estimator=est, mu=0.0, m=1.0)
# oc_tuned = OcDecomposition(base_estimator=bc)
# oc_tuned.fit(x_train, y_train, mus=mus, ms=ms)
# accuracy_tuned = oc_tuned.accuracy(x_test, y_test, mus=mus,
# ms=ms)
# accuracies_tuned[mc * n_folds + test_fold] = accuracy_tuned
# diary.add_entry('validation', ['dataset', name,
# 'method', 'BC-tuned',
# 'mc', mc,
# 'test_fold', test_fold,
# 'acc', accuracy_tuned])
# df = df.append_rows([[name, 'BC-tuned', mc, test_fold,
# accuracy_tuned]])
export_summary(df, diary)
try:
perms = [
perm['name']
for perm in vk.method('groups.getTokenPermissions')['permissions']
]
if 'manage' not in perms or 'messages' not in perms:
call_exit('У ключа недостаточно прав')
except ApiError:
call_exit('Неверный ключ доступа')
try:
vk.method('groups.getOnlineStatus', {'group_id': parser['Vk']['group_id']})
except Exception:
call_exit('В настройках группы отключены сообщения или неверный id группы')
d = Diary(parser['Diary']['diary_login'], parser['Diary']['diary_password'],
session)
try:
d.auth()
except ValueError:
call_exit('Неверный логин или пароль')
except requests.exceptions.HTTPError:
call_exit('Электронный дневник не работает. Попробуйте запустить позже')
payload = {
'group_id': parser['Vk']['group_id'],
'enabled': 1,
'api_version': '5.92',
'message_new': 1
}
try:
vk.method('groups.setLongPollSettings', payload)
