def test_update_unfit_groups_with_crossgroup_dist_03(self):
# load test data
user_ids, user_profile_df, user_connection_df = load_sample_test_data()
# define test data
fit_group = {0:[], 1:[]}
fit_pvals = {0:[], 1:[]}
unfit_group = {0: ['d', 'e'], 1: ['a', 'b', 'c']}
buffer_group = []
dist_metrics = {0: [1, 1, 1], 1: [0, 0, 1]}
new_fit_group, new_fit_pvals, buffer_group = _update_unfit_groups_with_crossgroup_dist(dist_metrics,
fit_group,
fit_pvals,
unfit_group,
buffer_group,
user_ids,
user_profile_df,
user_connection_df,
ks_alpha=0.05)
print("--- _update_unfit_groups_with_crossgroup_dist (with reverse dist metrics)---")
print("1st unfit_group: {}".format(unfit_group[0]))
print("2nd unfit_group: {}".format(unfit_group[1]))
print("new 1st group's distance metrics: {}".format(dist_metrics[0]))
print("new 2nd group's distance metrics: {}".format(dist_metrics[1]))
print("new 1st fit_group: {}".format(new_fit_group[0]))
print("new 2nd fit_group: {}".format(new_fit_group[1]))
print("new 1st fit_pval: {}".format(new_fit_pvals[0]))
print("new 2nd fit_pval: {}".format(new_fit_pvals[1]))
print("buffer_group: {}".format(buffer_group))
def test_single_run_05(self):
user_ids, user_profiles, user_connections = load_sample_test_data()
# definte test data
fit_group = {0: [], 1: ['a', 'd', 'e']}
fit_pvals = {0: [], 1: [1, 1]}
dist_metrics = {0: [1, 1, 1], 1: [1, 1, 1]}
buffer_group = []
try:
single_run_res = _groupwise_dist_learning_single_run(
dist_metrics,
fit_group,
fit_pvals,
buffer_group,
user_ids,
user_profiles,
user_connections,
ks_alpha=0.05,
min_group_size=1,
verbose=True)
# test internal mehtod regardin validating input data
# it should capture the illegal input
is_ok = False
except:
is_ok = True
self.assertTrue(is_ok)
def test_update_fit_group_with_groupwise_dist_01(self):
""" test with generic distance metrics """
# load test data
user_ids, user_profile_df, user_connection_df = load_sample_test_data()
# definte test data
fit_group = {0: ['a', 'b', 'c'], 1: ['d', 'e']}
fit_pvals = {0: [1, 1, 1], 1: [1, 1]}
dist_metrics = {0: [1, 1, 1], 1: [1, 1, 1]}
dist_memory_container = [{}, {}]
new_fit_group, new_fit_pvals, unfit_group = _update_fit_group_with_groupwise_dist(
dist_metrics,
fit_group,
fit_pvals,
user_ids,
user_profile_df,
user_connection_df,
dist_memory_container,
is_directed=False,
ks_alpha=0.05)
print(
"--- update_fit_group_with_groupwise (with: generic distance metrics) ---"
)
print("new 1st fit_group: {}".format(new_fit_group[0]))
print("new 2nd fit_group: {}".format(new_fit_group[1]))
print("new 1st fit_pval: {}".format(new_fit_pvals[0]))
print("new 2nd fit_pval: {}".format(new_fit_pvals[1]))
is_ok = True
self.assertTrue(is_ok, True)
def test_update_buffer_group(self):
""" test _update_buffer_group """
# load test data
user_ids, user_profile_df, user_connection_df = load_sample_test_data()
# define test data
fit_group = {0: [], 1: []}
fit_pvals = {0: [], 1: []}
buffer_group = user_ids
new_dist_metrics = {0: [0, 1, 0], 1: [0, 0, 1]}
dist_memory_container = [{}, {}]
new_fit_group, new_fit_pvals, unfit_group = _update_buffer_group(
new_dist_metrics,
fit_group,
fit_pvals,
buffer_group,
user_ids,
user_profile_df,
user_connection_df,
dist_memory_container,
ks_alpha=0.5)
print("--- _update_buffer_group (with empty fit_group)---")
print("new 1st fit_group: {}".format(new_fit_group[0]))
print("new 2nd fit_group: {}".format(new_fit_group[1]))
is_ok = True
self.assertTrue(is_ok, True)
def test_single_run_03(self):
user_ids, user_profiles, user_connections = load_sample_test_data()
# definte test data
fit_group = {0: [], 1: ['a', 'd', 'e', 'b', 'c']}
fit_pvals = {0: [], 1: [1, 1, 1, 1, 1]}
dist_metrics = {0: [1, 1, 1], 1: [1, 1, 1]}
buffer_group = []
single_run_res = _groupwise_dist_learning_single_run(dist_metrics,
fit_group,
fit_pvals,
buffer_group,
user_ids,
user_profiles,
user_connections,
ks_alpha=0.05,
min_group_size=1,
verbose=True)
new_dist_metrics, new_fit_group, new_fit_pvals, new_buffer_group = single_run_res
print("--- single run test ---")
print("1st new_dist_metrics's distance metrics: {}".format(
new_dist_metrics[0]))
print("1nd new_dist_metrics's distance metrics: {}".format(
new_dist_metrics[1]))
print("new 1st fit_group: {}".format(new_fit_group[0]))
print("new 2nd fit_group: {}".format(new_fit_group[1]))
print("new 1st fit_pval: {}".format(new_fit_pvals[0]))
print("new 2nd fit_pval: {}".format(new_fit_pvals[1]))
print("buffer_group: {}".format(new_buffer_group))
def test_update_groupwise_dist(self):
user_ids, user_profile_df, user_connection_df = load_sample_test_data()
# define test data
fit_group = {0: ['a', 'b', 'c'], 1: ['d', 'e']}
dist_metrics = {0: [1, 1, 1], 1: [1, 1, 1]}
new_dist_metrics = _update_groupwise_dist(dist_metrics,
fit_group,
user_ids,
user_profile_df,
user_connection_df,
min_group_size=1)
print(
"--- test_update_groupwise_dist (with generic metrics as inputs) ---"
)
print("1st group's old distance weights: {}".format(dist_metrics[0]))
print("2nd group's old distance weights: {}".format(dist_metrics[1]))
print("1st group's new distance weights: {}".format(
new_dist_metrics[0]))
print("2nd group's new distance weights: {}".format(
new_dist_metrics[1]))
is_ok = True
self.assertTrue(is_ok, True)
def test_update_fit_group_with_groupwise_dist_02(self):
""" test with generic distance metrics """
# load test data
user_ids, user_profile_df, user_connection_df = load_sample_test_data()
# definte test data
fit_group = {0: ['a', 'b', 'c'], 1: ['d', 'e']}
fit_pvals = {0: [1, 1, 1], 1: [1, 1]}
new_dist_metrics = {0: [0, 1, 0], 1: [0, 0, 1]}
new_fit_group, new_fit_pvals, unfit_group = _update_fit_group_with_groupwise_dist(new_dist_metrics,
fit_group,
fit_pvals,
user_ids,
user_profile_df,
user_connection_df,
ks_alpha=0.05)
print("--- update_fit_group_with_groupwise (with: learned distance metrics) ---")
print("new 1st fit_group: {}".format(new_fit_group[0]))
print("new 2nd fit_group: {}".format(new_fit_group[1]))
print("new 1st fit_pval: {}".format(new_fit_pvals[0]))
print("new 2nd fit_pval: {}".format(new_fit_pvals[1]))
is_ok = True
self.assertTrue(is_ok, True)
def test_learner_01(self):
user_ids, user_profiles, user_connections = load_sample_test_data()
best_pack = groupwise_dist_learning(user_ids,
user_profiles,
user_connections,
n_group=2,
max_iter=20,
max_nogain_streak=5,
min_group_size=1,
ks_alpha=0.1,
init='zipf',
verbose=True,
C=0.1)
knowledge_pack, best_score = best_pack
new_dist_metrics, new_fit_group, new_buffer_group = knowledge_pack
print("--- leaner test (n_group=2) ---")
print("1st new_dist_metrics's distance metrics: {}".format(
new_dist_metrics[0]))
print("1nd new_dist_metrics's distance metrics: {}".format(
new_dist_metrics[1]))
print("new 1st fit_group: {}".format(new_fit_group[0]))
print("new 2nd fit_group: {}".format(new_fit_group[1]))
print("buffer_group: {}".format(new_buffer_group))
def test_learner_class_init_zipf(self):
user_ids, user_profiles, user_connections = load_sample_test_data()
gwd_learner = GroupwiseDistLearner(n_group=2, min_group_size=1, init="zipf", max_iter=10, verbose=True)
gwd_learner.fit(user_ids, user_profiles, user_connections)
print("--- learner class (n_group=2) with init='zipf' ---")
print("best score: {}".format(gwd_learner.get_score()))
def test_validate_user_informationg_with_good_data(self):
user_ids, user_profile_df, user_connection_df = load_sample_test_data()
try:
_validate_user_information(user_ids = user_ids,
user_profiles = user_profile_df,
user_connections = user_connection_df)
is_ok = True
except:
is_ok = False
self.assertTrue(is_ok)
def test_validate_user_informationg_with_good_data(self):
user_ids, user_profile_df, user_connection_df = load_sample_test_data()
try:
_validate_user_information(user_ids=user_ids,
user_profiles=user_profile_df,
user_connections=user_connection_df)
is_ok = True
except:
is_ok = False
self.assertTrue(is_ok)
def test_learner_class_init_zipf(self):
user_ids, user_profiles, user_connections = load_sample_test_data()
gwd_learner = GroupwiseDistLearner(n_group=2,
min_group_size=1,
init="zipf",
max_iter=10,
verbose=True)
gwd_learner.fit(user_ids, user_profiles, user_connections)
print("--- learner class (n_group=2) with init='zipf' ---")
print("best score: {}".format(gwd_learner.get_score()))
def test_validate_user_informationg_with_bad_data(self):
user_ids, user_profile_df, user_connection_df = load_sample_test_data()
try:
# remove first user_id's record
_validate_user_information(user_ids = user_ids[1:],
user_profiles = user_profile_df,
user_connections = user_connection_df)
is_ok = False
except:
is_ok = True
self.assertTrue(is_ok)
def test_validate_user_informationg_with_bad_data(self):
user_ids, user_profile_df, user_connection_df = load_sample_test_data()
try:
# remove first user_id's record
_validate_user_information(user_ids=user_ids[1:],
user_profiles=user_profile_df,
user_connections=user_connection_df)
is_ok = False
except:
is_ok = True
self.assertTrue(is_ok)
def test_learner_02(self):
user_ids, user_profiles, user_connections = load_sample_test_data()
best_pack = groupwise_dist_learning(user_ids, user_profiles, user_connections, n_group=1,
max_iter=20, max_nogain_streak=5, min_group_size=1, ks_alpha=0.1,
init='zipf', verbose=True, C=0.1)
knowledge_pack, best_score = best_pack
new_dist_metrics, new_fit_group, new_buffer_group = knowledge_pack
print("--- leaner test (n_group=1) ---")
print("1st new_dist_metrics's distance metrics: {}".format(new_dist_metrics[0]))
print("new 1st fit_group: {}".format(new_fit_group[0]))
print("buffer_group: {}".format(new_buffer_group))
def test_update_buffer_group(self):
""" test _update_buffer_group """
# load test data
user_ids, user_profile_df, user_connection_df = load_sample_test_data()
# define test data
fit_group = {0:[], 1:[]}
fit_pvals = {0:[], 1:[]}
buffer_group = user_ids
new_dist_metrics = {0: [0, 1, 0], 1: [0, 0, 1]}
new_fit_group, new_fit_pvals, unfit_group = _update_buffer_group(new_dist_metrics, fit_group, fit_pvals,
buffer_group, user_ids, user_profile_df,
user_connection_df, ks_alpha=0.5)
print("--- _update_buffer_group (with empty fit_group)---")
print("new 1st fit_group: {}".format(new_fit_group[0]))
print("new 2nd fit_group: {}".format(new_fit_group[1]))
is_ok = True
self.assertTrue(is_ok, True)
def test_single_run_05(self):
user_ids, user_profiles, user_connections = load_sample_test_data()
# definte test data
fit_group = {0: [], 1: ['a', 'd', 'e']}
fit_pvals = {0: [], 1: [1, 1]}
dist_metrics = {0: [1, 1, 1], 1: [1, 1, 1]}
buffer_group = []
try:
single_run_res = _groupwise_dist_learning_single_run(dist_metrics, fit_group, fit_pvals, buffer_group,
user_ids, user_profiles, user_connections,
ks_alpha=0.05, min_group_size=1, verbose=True)
# test internal mehtod regardin validating input data
# it should capture the illegal input
is_ok = False
except:
is_ok = True
self.assertTrue(is_ok)
def test_update_groupwise_dist(self):
user_ids, user_profile_df, user_connection_df = load_sample_test_data()
# define test data
fit_group = {0: ['a', 'b', 'c'], 1: ['d', 'e']}
dist_metrics = {0: [1, 1, 1], 1: [1, 1, 1]}
new_dist_metrics = _update_groupwise_dist(dist_metrics, fit_group,
user_ids, user_profile_df, user_connection_df,
min_group_size=1)
print("--- test_update_groupwise_dist (with generic metrics as inputs) ---")
print("1st group's old distance weights: {}".format(dist_metrics[0]))
print("2nd group's old distance weights: {}".format(dist_metrics[1]))
print("1st group's new distance weights: {}".format(new_dist_metrics[0]))
print("2nd group's new distance weights: {}".format(new_dist_metrics[1]))
is_ok = True
self.assertTrue(is_ok, True)
def test_update_unfit_groups_with_crossgroup_dist_02(self):
# load test data
user_ids, user_profile_df, user_connection_df = load_sample_test_data()
# define test data
fit_group = {0: [], 1: []}
fit_pvals = {0: [], 1: []}
unfit_group = {0: ['d', 'e'], 1: ['a', 'b', 'c']}
buffer_group = []
dist_metrics = {0: [1, 0, 1], 1: [1, 1, 0]}
dist_memory_container = [{}, {}]
new_fit_group, new_fit_pvals, buffer_group = _update_unfit_groups_with_crossgroup_dist(
dist_metrics,
fit_group,
fit_pvals,
unfit_group,
buffer_group,
user_ids,
user_profile_df,
user_connection_df,
dist_memory_container,
ks_alpha=0.05)
print(
"--- _update_unfit_groups_with_crossgroup_dist (with modified dist metrics)---"
)
print("1st unfit_group: {}".format(unfit_group[0]))
print("2nd unfit_group: {}".format(unfit_group[1]))
print("new 1st group's distance metrics: {}".format(dist_metrics[0]))
print("new 2nd group's distance metrics: {}".format(dist_metrics[1]))
print("new 1st fit_group: {}".format(new_fit_group[0]))
print("new 2nd fit_group: {}".format(new_fit_group[1]))
print("new 1st fit_pval: {}".format(new_fit_pvals[0]))
print("new 2nd fit_pval: {}".format(new_fit_pvals[1]))
print("buffer_group: {}".format(buffer_group))
is_ok = True
self.assertTrue(is_ok, True)
def test_single_run_03(self):
user_ids, user_profiles, user_connections = load_sample_test_data()
# definte test data
fit_group = {0: [], 1: ['a', 'd', 'e', 'b', 'c']}
fit_pvals = {0: [], 1: [1, 1, 1, 1, 1]}
dist_metrics = {0: [1, 1, 1], 1: [1, 1, 1]}
buffer_group = []
single_run_res = _groupwise_dist_learning_single_run(dist_metrics, fit_group, fit_pvals, buffer_group,
user_ids, user_profiles, user_connections,
ks_alpha=0.05, min_group_size=1, verbose=True)
new_dist_metrics, new_fit_group, new_fit_pvals, new_buffer_group = single_run_res
print("--- single run test ---")
print("1st new_dist_metrics's distance metrics: {}".format(new_dist_metrics[0]))
print("1nd new_dist_metrics's distance metrics: {}".format(new_dist_metrics[1]))
print("new 1st fit_group: {}".format(new_fit_group[0]))
print("new 2nd fit_group: {}".format(new_fit_group[1]))
print("new 1st fit_pval: {}".format(new_fit_pvals[0]))
print("new 2nd fit_pval: {}".format(new_fit_pvals[1]))
print("buffer_group: {}".format(new_buffer_group))
