Your IP : 216.73.217.13
from collections import defaultdict
from _library import profiling_utils
class ProfilingUsers:
def __init__(self, orders):
self.orders = orders
self.orders.sort_values(by = 'customer_id', inplace = True)
def mine_orders(self):
# Initialize user profiles
self.user_profiles = defaultdict(lambda: defaultdict(set))
# Mine the orders
miner_func = lambda df_row: profiling_utils.retrive_userCategories(df_row, self.user_profiles)
self.orders.apply(miner_func, axis = 1)
customers_consents = self.orders.drop_duplicates(subset=['customer_id'])
customers_consents = customers_consents[['customer_id','consenso_dati_particolari']]
for user_profile_key in self.user_profiles:
# ATTRIBUTE: Weight
min_weight = int(min(self.user_profiles[user_profile_key]['product_weights']))
max_weight = int(max(self.user_profiles[user_profile_key]['product_weights']))
self.user_profiles[user_profile_key]['product_weights'] = {'min': min_weight, 'max': max_weight}
user_products_info = self.orders[self.orders['SKU'].isin(self.user_profiles[user_profile_key]['unique_products'])].drop_duplicates(subset=['SKU'])
# ATTRIBUTE: is adult
self.user_profiles[user_profile_key]['adult'] = 1 if any(p == 1 for p in user_products_info['consenso_adult'].values) else 0
# ATTRIBUTE: purchased biologic, vegan, biodinamic products as percentage
#-1 = no consenso per dati particolari
self.user_profiles[user_profile_key]['bio_percentage'] = -1
self.user_profiles[user_profile_key]['biodinamic_percentage'] = -1
self.user_profiles[user_profile_key]['vegan_percentage'] = -1
self.user_profiles[user_profile_key]['gluten_free_percentage'] = -1
if(customers_consents.loc[customers_consents['customer_id'] == 1]['consenso_dati_particolari'].values[0] is not None):
user_purchased_bio_percentage = sum(user_products_info['biologic']) / len(user_products_info['biologic'])
user_purchased_vegan_percentage = sum(user_products_info['vegan']) / len(user_products_info['vegan'])
user_purchased_biodinamic_percentage = sum(user_products_info['biodinamic']) / len(user_products_info['biodinamic'])
user_purchased_gluten_free = sum(user_products_info['gluten_free']) / len(user_products_info['gluten_free'])
self.user_profiles[user_profile_key]['bio_percentage'] = user_purchased_bio_percentage
self.user_profiles[user_profile_key]['biodinamic_percentage'] = user_purchased_biodinamic_percentage
self.user_profiles[user_profile_key]['vegan_percentage'] = user_purchased_vegan_percentage
self.user_profiles[user_profile_key]['gluten_free_percentage'] = user_purchased_gluten_free
# Unknown customer id --> no user linked to the order
if -1 in self.user_profiles.keys():
self.user_profiles.pop(-1)
return self.user_profiles