def setUp(self):
if not getattr(TestRegressions, '_dir_changed', False):
abspath = os.path.abspath(__file__)
name = abspath.index(os.path.basename(__file__))
abspath = abspath[:name]
os.chdir(abspath)
TestRegressions._dir_changed = True
self.empty_user = bc.User()
self.empty_user.attributes['empty'] = True
self.sample_user = bc.tests.generate_user.sample_user()
self.network_ego = bc.read_csv(
'ego', 'samples/network', 'samples/towers.csv',
attributes_path='samples/attributes', network=True, describe=False)
# Manual users
self.user_a = bc.read_csv(
'A', 'samples/manual', 'samples/towers.csv',
recharges_path='samples/manual/recharges', network=False,
describe=False)
self.user_a_network = bc.read_csv(
'A', 'samples/manual', 'samples/towers.csv',
attributes_path='samples/attributes', network=True, describe=False)
self.user_a_orange = bc.io.read_orange(
'A_orange', 'samples/manual', 'sample/towers.csv',
recharges_path='samples/manual/recharges', network=False,
describe=False)
self.user_a_orange_network = bc.io.read_orange(
'A_orange', 'samples/manual', network=True,
attributes_path='samples/attributes', describe=False)
def main():
# read in attributes for all user
if args.attributes_input_file:
user_attributes = read_user_attributes(args.attributes_input_file)
indicators = []
for dir in os.listdir(args.cdrs_input_dir):
cdrs_dir = os.path.join(args.cdrs_input_dir, dir)
if not os.path.isdir(cdrs_dir):
continue
for file in os.listdir(cdrs_dir):
cdr_input_file = os.path.join(cdrs_dir, file)
user_name = file
# should we add atributes to indicators?
if args.attributes_input_file:
# find the attributes for this user and write them to an in-memory file
if user_name not in user_attributes:
sys.exit('Could not find attributes for user ' + user_name)
gender = user_attributes[user_name][0]
age = user_attributes[user_name][1] if user_attributes[
user_name][1] else "None"
attributes_file = tempfile.NamedTemporaryFile()
attributes_file.write("key,value\n")
attributes_file.write("gender," + str(gender) + "\n")
attributes_file.write("age," + str(age))
# don't forget to seek to beginning so that what you wrote can be read!
attributes_file.seek(0)
user = bandicoot.read_csv(cdr_input_file,
args.antennas_input_file,
attributes_path=attributes_file.name,
describe=False)
attributes_file.close()
else:
user = bandicoot.read_csv(cdr_input_file,
args.antennas_input_file,
describe=False)
# set the username to filename
user.name = user_name
# Extended summary also computes skewness and kurtosis
user_indicators = bandicoot.utils.all(user,
weekly=True,
summary='extended',
attributes=True)
indicators.append(user_indicators)
bandicoot.io.to_csv(indicators, args.indicators_output)
if args.delete_bad_columns:
delete_bad_columns(args.indicators_output)
def load_and_compute(f):
user_id = os.path.basename(f)[:-4]
try:
B = bc.read_csv(user_id, records_path, antenna_file, describe=False)
metrics_dic = bc.utils.all(B)
except Exception as e:
metrics_dic = {'name': user_id, 'error': True}
return metrics_dic
def test_read_csv_attributes(self):
user = bc.read_csv("u_test2", "samples", attributes_path="samples/attributes", describe=False)
self.assertEqual(user.attributes, {
'gender': 'male',
'age': '42',
'is_subscriber': 'True',
'individual_id': '7atr8f53fg41'
})
def load_and_compute(f):
user_id = os.path.basename(f)[:-4]
try:
B = bc.read_csv(user_id, records_path, antenna_file, describe=False)
metrics_dic = bc.utils.all(B)
except Exception as e:
metrics_dic = {'name': user_id, 'error': True}
return metrics_dic
def setUp(self):
if not getattr(TestRegressions, '_dir_changed', False):
abspath = os.path.abspath(__file__)
name = abspath.index(os.path.basename(__file__))
abspath = abspath[:name]
os.chdir(abspath)
TestRegressions._dir_changed = True
self.empty_user = bc.User()
self.empty_user.attributes['empty'] = True
self.sample_user = bc.tests.generate_user.sample_user()
self.network_ego = bc.read_csv('ego',
'samples/network',
'samples/towers.csv',
attributes_path='samples/attributes',
network=True,
warnings=False,
describe=False)
# Manual users
self.user_a = bc.read_csv('A',
'samples/manual',
'samples/towers.csv',
network=False,
warnings=False,
describe=False)
self.user_a_network = bc.read_csv('A',
'samples/manual',
'samples/towers.csv',
attributes_path='samples/attributes',
network=True,
warnings=False,
describe=False)
self.user_a_orange = bc.io.read_orange('A_orange',
'samples/manual',
network=False,
warnings=False,
describe=False)
self.user_a_orange_network = bc.io.read_orange(
'A_orange',
'samples/manual',
network=True,
attributes_path='samples/attributes',
warnings=False,
describe=False)
def test_read_csv_no_position(self):
user = bc.read_csv("u_test_no_position", "samples", describe=False)
self.assertEqual(user.records[1],
Record(interaction='call',
direction='in',
correspondent_id='770000001',
datetime=datetime.datetime(2013, 12, 16, 5, 39, 30),
call_duration=0,
position=Position()))
def test_read_csv_attributes(self):
user = bc.read_csv("u_test2", "samples",
attributes_path="samples/attributes", describe=False)
self.assertEqual(user.attributes, {
'gender': 'male',
'age': '42',
'is_subscriber': 'True',
'individual_id': '7atr8f53fg41'
})
def test_read_csv_no_position(self):
user = bc.read_csv("u_test_no_position", "samples", describe=False)
self.assertEqual(
user.records[1],
Record(interaction='call',
direction='in',
correspondent_id='770000001',
datetime=datetime.datetime(2013, 12, 16, 5, 39, 30),
duration=0,
position=Position()))
def test_read_csv_antenna_id_no_places(self):
user = bc.read_csv("u_test_antennas", "samples", describe=False)
self.assertEqual(user.records[1],
Record(interaction='call',
direction='in',
correspondent_id='770000001',
datetime=datetime.datetime(2013, 12, 16, 5, 39, 30),
call_duration=0,
position=Position('13084', None)))
self.assertEqual(bc.spatial.radius_of_gyration(user).values(), [None] * 2)
def test_read_csv_antenna_id(self):
user = bc.read_csv("u_test_antennas", "samples", antennas_path="samples/towers.csv", describe=False)
self.assertEqual(user.records[1],
Record(interaction='call',
direction='in',
correspondent_id='770000001',
datetime=datetime.datetime(2013, 12, 16, 5, 39, 30),
call_duration=0,
position=Position('13084', None)))
radius = bc.spatial.radius_of_gyration(user)['mean']
self.assertGreater(radius, 0)
def test_read_csv_antenna_id_no_places(self):
user = bc.read_csv("u_test_antennas", "samples", describe=False)
self.assertEqual(user.records[1],
Record(interaction='call',
direction='in',
correspondent_id='770000001',
datetime=datetime.datetime(2013, 12, 16, 5, 39, 30),
duration=0,
position=Position('13084', None)))
result = {'allweek': {'allday': None}}
self.assertEqual(bc.spatial.radius_of_gyration(user, groupby=None), result)
def network_sampling(n, filename, directory=None, snowball=False, user=None):
"""
Selects a few users and exports a CSV of indicators for them.
TODO: Returns the network/graph between the selected users.
Parameters
----------
n : int
Number of users to select.
filename : string
File to export to.
directory: string
Directory to select users from if using the default random selection.
Selection options
-----------------
random (default): selects n users at random
snowball: starts from a specified user, iterates over neighbors, and does a BFS until n neighbors are reached
"""
if snowball:
if user is None:
raise ValueError(
"Must specify a starting user from whom to initiate the snowball"
)
else:
users, agenda = [user], [user]
while len(agenda) > 0:
parent = agenda.pop()
dealphebetized_network = sorted(parent.network.items(),
key=lambda k: random.random())
for neighbor in dealphebetized_network:
if neighbor[1] not in users and neighbor[
1] is not None and len(users) < n:
users.append(neighbor[1])
if neighbor[1].network:
agenda.push(neighbor[1])
else:
files = [
x for x in os.listdir(directory)
if os.path.isfile(os.path.join(directory, x))
]
shuffled_files = sorted(files, key=lambda k: random.random())
user_names = shuffled_files[:n]
users = [bc.read_csv(u[:-4], directory) for u in user_names]
if len(users) < n:
raise ValueError(
"Specified more users than records that exist, only {} records available"
.format(len(users)))
bc.to_csv([bc.utils.all(u) for u in users], filename)
def test_read_csv_antenna_id_no_places(self):
user = bc.read_csv("u_test_antennas", "samples", describe=False)
self.assertEqual(user.records[1],
Record(interaction='call',
direction='in',
correspondent_id='770000001',
datetime=dt(2013, 12, 16, 5, 39, 30),
call_duration=0,
position=Position('13084', None)))
result = {'allweek': {'allday': None}}
radius = bc.spatial.radius_of_gyration(user, groupby=None)
self.assertEqual(radius, result)
def test_read_csv_antenna_id_no_places(self):
user = bc.read_csv("u_test_antennas", "samples", describe=False)
self.assertEqual(
user.records[1],
Record(interaction='call',
direction='in',
correspondent_id='770000001',
datetime=datetime.datetime(2013, 12, 16, 5, 39, 30),
call_duration=0,
position=Position('13084', None)))
self.assertEqual(
bc.spatial.radius_of_gyration(user).values(), [None] * 2)
def test_read_csv_antenna_id(self):
user = bc.read_csv("u_test_antennas", "samples",
antennas_path="samples/towers.csv", describe=False)
self.assertEqual(user.records[1],
Record(interaction='call',
direction='in',
correspondent_id='770000001',
datetime=dt(2013, 12, 16, 5, 39, 30),
call_duration=0,
position=Position('13084', None)))
radius = bc.spatial.radius_of_gyration(user, groupby=None)
self.assertGreater(radius['allweek']['allday'], 0)
def metrics():
records_path = 'records/'
antenna_file = 'antennas.csv'
indicators = []
for f in glob.glob(records_path + '*.csv'):
user_id = os.path.basename(f)[:-4]
try:
B = bc.read_csv(user_id, records_path, antenna_file, describe=False)
metrics_dict = bc.utils.all(B)
except Exception as e:
metrics_dict = {'name': user_id, 'error': True}
indicators.append(metrics_dict)
bc.io.to_csv(indicators, 'bandicoot_indicators_full.csv')
def network_sampling(n, filename, directory=None, snowball=False, user=None):
"""
Selects a few users and exports a CSV of indicators for them.
TODO: Returns the network/graph between the selected users.
Parameters
----------
n : int
Number of users to select.
filename : string
File to export to.
directory: string
Directory to select users from if using the default random selection.
Selection options
-----------------
random (default): selects n users at random
snowball: starts from a specified user, iterates over neighbors, and does a BFS until n neighbors are reached
"""
if snowball:
if user is None:
raise ValueError("Must specify a starting user from whom to initiate the snowball")
else:
users, agenda = [user], [user]
while len(agenda) > 0:
parent = agenda.pop()
dealphebetized_network = sorted(parent.network.items(), key=lambda k: random.random())
for neighbor in dealphebetized_network:
if neighbor[1] not in users and neighbor[1] is not None and len(users) < n:
users.append(neighbor[1])
if neighbor[1].network:
agenda.push(neighbor[1])
else:
files = [x for x in os.listdir(directory) if os.path.isfile(os.path.join(directory, x))]
shuffled_files = sorted(files, key=lambda k: random.random())
user_names = shuffled_files[:n]
users = [bc.read_csv(u[:-4], directory) for u in user_names]
if len(users) < n:
raise ValueError("Specified more users than records that exist, only {} records available".format(len(users)))
bc.to_csv([bc.utils.all(u) for u in users], filename)
def load_and_compute(user_id, attributes):
"""
Bandicoot helper function with inputs based on the predefined attributes
in the attributes class.
Inputs
------
user_id : ID of user on which bandicoot features are being calculated.
attributes : Attributes class with specific options for current run.
Output
------
Dictionary of calculated bandicoot indicators.
"""
try:
# create user object &
# ignore massive warnings output for better speed
B = bc.read_csv(user_id=user_id,
records_path=attributes.bandicoot_path,
antennas_path=attributes.bandicoot_path+'antennas.csv',
describe=False,
warnings=False)
# change weekend days and nocturnal time
B.weekend = attributes.weekend_days
B.night_start = datetime.time(attributes.noct_time['begin'])
B.night_end = datetime.time(attributes.noct_time['end'])
# group by month (aka insert user data in chunks of months)
# & calculate for weekend and workweek separately
# & calculate for day and night separately:
metrics_dic = bc.utils.all(B,
groupby=None,
split_week=True,
split_day=True,
summary='extended')
except Exception as e:
metrics_dic = {'name': user_id, 'error': True}
return metrics_dic
def test_read_csv(self):
user = bc.read_csv("u_test2", "samples", describe=False)
self.assertEqual(len(user.records), 500)
def setUp(self):
self.user = bc.read_csv('user_ignored', 'samples')
import bandicoot as bc
import pprint
u = bc.read_csv("sanjay", "C:\\Users\\Sahil\\Desktop")
rec = u.records
#other features that can be included - call duration... the people who are being called -- the number of times they have been called in total
t = []
c = 0
#print(rec[1].datetime.timestamp())
for i in rec:
#format i.datetime
t.append(i.datetime.timestamp())
c += 1
def grouper(iterable):
prev = None
group = []
for item in iterable:
if not prev or item - prev <= 1000:
group.append(item)
else:
yield group
group = [item]
prev = item
if group:
yield group
di = dict(enumerate(grouper(t), 1))
import sys
sys.path.append("../")
import bandicoot as bc
import glob
import os
records_path = 'users_bandicoot/'
antenna_file = 'towers.csv'
indicators = []
for f in glob.glob(records_path + '*.csv'):
user_id = os.path.basename(f)[:-4]
try:
B = bc.read_csv(user_id, records_path, antenna_file, describe=False)
metrics_dict = bc.utils.all(B)
except Exception as e:
metrics_dic = {'name': user_id, 'error': True}
indicators.append(metrics_dict)
bc.io.to_csv(indicators, 'bandicoot_indicators_full.csv')
def setUp(self):
self.user = bc.read_csv('user_ignored', 'samples')
from bandicoot.helper.group import grouping
import bandicoot as bc
# Loading a User
U = bc.read_csv('ego', 'data/', 'data/antennas.csv')
#######################
# Export visulization #
#######################
bc.visualization.export(U, 'my-viz-path')
#########################################
# Run individual and spatial indicators #
#########################################
bc.individual.percent_initiated_conversations(U)
bc.spatial.number_of_antennas(U)
bc.spatial.radius_of_gyration(U)
######################################
# Group indicators by weeks or month #
######################################
# The groupby keyword controls the aggregation:
# - groupby='week' to divide by week (by default),
# - groupby='month' to divide by month,
def test_read_csv_with_recharges(self):
user = bc.read_csv("A",
"samples/manual",
describe=False,
recharges_path="samples/manual/recharges")
self.assertEqual(len(user.recharges), 5)
def test_read_csv_with_recharges(self):
user = bc.read_csv("A", "samples/manual", describe=False,
recharges_path="samples/manual/recharges")
self.assertEqual(len(user.recharges), 5)
def test_read_csv(self):
user = bc.read_csv("u_test2", "samples", describe=False)
self.assertEqual(len(user.records), 500)
