def read_bidders(filename): with open('clean_%s' % filename, 'wb') as cbfile: fieldnames = [ 'bidder_id', 'pmt_type', 'pmt_account', 'addr_1', 'addr_2' ] if 'train' in filename: fieldnames.append('outcome') writer = csv.DictWriter(cbfile, fieldnames=fieldnames) writer.writeheader() with open(filename, 'rb') as bidsfile: reader = csv.reader(bidsfile, delimiter=',', quotechar='|') skip(reader, 1) for row in reader: b = Bidder(row[0], row[1], row[2], row[3] if len(row) > 3 else None) bidders[b.bidder_id] = b mrow = { 'bidder_id': b.bidder_id, 'pmt_type': b.get_payment_type(), 'pmt_account': b.get_payment_acct(), 'addr_1': b.get_addr_1(), 'addr_2': b.get_addr_2(), } if 'train' in filename: mrow['outcome'] = b.outcome writer.writerow(mrow)
def read_bidders(filename): with open(filename, 'rb') as bidsfile: reader = csv.reader(bidsfile, delimiter=',', quotechar='|') skip(reader, 1) for row in reader: b = Bidder(row[0], row[1], row[2], row[3] if len(row) > 3 else None) #if b.bidder_id in ['f5b2bbad20d1d7ded3ed960393bec0f40u6hn', 'e90e4701234b13b7a233a86967436806wqqw4']: # b.outcome = 1 add_bidder(b) shared.addr_1[b.get_addr_1()].add(b.bidder_id) shared.addr_2[b.get_addr_2()].add(b.bidder_id) shared.pmt_type[b.get_payment_type()].add(b.bidder_id) shared.pmt_accnt[b.get_payment_acct()].add(b.bidder_id) print "%s data done" % filename rc = sum([1 for i in shared.bidders.values() if utils.is_robot(i)]) hc = sum([1 for i in shared.bidders.values() if utils.is_human(i)]) uc = sum([1 for i in shared.bidders.values() if i.outcome is None]) print "Found %s robots, %s humans, %s unknown" % (rc, hc, uc)
def read_bidders(filename): with open('clean_%s' % filename, 'wb') as cbfile: fieldnames = ['bidder_id', 'pmt_type', 'pmt_account', 'addr_1', 'addr_2'] if 'train' in filename: fieldnames.append('outcome') writer = csv.DictWriter(cbfile, fieldnames=fieldnames) writer.writeheader() with open(filename, 'rb') as bidsfile: reader = csv.reader(bidsfile, delimiter=',', quotechar='|') skip(reader, 1) for row in reader: b = Bidder(row[0], row[1], row[2], row[3] if len(row) > 3 else None) bidders[b.bidder_id] = b mrow = { 'bidder_id': b.bidder_id, 'pmt_type': b.get_payment_type(), 'pmt_account': b.get_payment_acct(), 'addr_1': b.get_addr_1(), 'addr_2': b.get_addr_2(), } if 'train' in filename: mrow['outcome'] = b.outcome writer.writerow(mrow)