def main(self):
parser = argparse.ArgumentParser(description='modeling customer profile data with text')
parser.add_argument('-g', action='store_true', help="generate customer and behavioral profile data")
parser.add_argument('-a', action='store_true', help="analyze behavioral profile data")
args = parser.parse_args()
try:
if args.g:
generate_profile_data(100,50,2,10,100)
elif args.a:
behavioral_profiles = Yamlator.load("behavioral_profiles.yaml")
customers = Yamlator.load("customers.yaml")
analyze(behavioral_profiles, customers)
except:
traceback.print_exc()
def main(self):
parser = argparse.ArgumentParser(
description='modeling customer profile data with text')
parser.add_argument(
'-g',
action='store_true',
help="generate customer and behavioral profile data")
parser.add_argument('-a',
action='store_true',
help="analyze behavioral profile data")
args = parser.parse_args()
try:
if args.g:
generate_profile_data(100, 50, 2, 10, 100)
elif args.a:
behavioral_profiles = Yamlator.load("behavioral_profiles.yaml")
customers = Yamlator.load("customers.yaml")
analyze(behavioral_profiles, customers)
except:
traceback.print_exc()
def generate_profile_data(num_customer_profiles, terms_per_profile,
min_products_purchased_per_customer,
max_products_purchased_per_customer,
num_products_per_behavioral_profile):
"""
Generates customer and behavioral profile data.
"""
#builds term/count map by genre
genre_freq_ass_map = dict()
print "building user product data set"
for f in glob.glob("seed/*.txt"):
genre = f.split("/")[1].split("-")[0]
if genre not in genre_freq_ass_map:
genre_freq_ass_map[genre] = FrequencyAssessor()
with open(f, "r") as current_file:
text = current_file.read()
tokens = nltk.word_tokenize(text)
for t in tokens:
if t not in string.punctuation and len(t) > 2:
if t.endswith("."):
t = t[:len(t) - 1]
genre_freq_ass_map[genre].update(t.lower())
#build genre>list of terms freq of occurance weighted.
genre_terms_list = dict()
for g in genre_freq_ass_map.keys():
genre_terms_list[g] = []
for t in genre_freq_ass_map[g].get_top_terms(max=500):
for i in range(t[1]):
genre_terms_list[g].append(t[0])
#build customer profiles
customer_profiles = []
for g in genre_terms_list.keys():
cid = 0
for c in range(num_customer_profiles):
product_descriptions = []
pid = 0
for p in range(
random.randint(min_products_purchased_per_customer,
max_products_purchased_per_customer)):
product_description = []
for i in range(terms_per_profile):
word_index = random.randint(0,
len(genre_terms_list[g]) - 1)
product_description.append(genre_terms_list[g][word_index])
z = 0
while z < 4:
#pop in some random words
ii = random.randint(0,
len(genre_terms_list.keys()) - 1)
random_genre = genre_terms_list.keys()[ii]
word_index = random.randint(
0,
len(genre_terms_list[random_genre]) - 1)
product_description.append(
genre_terms_list[random_genre][word_index])
z += 1
pid += 1
product_descriptions.append(
Product(pid, ' '.join(product_description)))
cid += 1
customer_profiles.append(Customer(g, cid, product_descriptions))
Yamlator.dump("customers.yaml", customer_profiles)
#build behavioral_profiles
behavioral_profiles = []
for g in genre_terms_list.keys():
product_descriptions = []
pid = 0
for p in range(num_products_per_behavioral_profile):
product_description = []
for i in range(terms_per_profile):
word_index = random.randint(0, len(genre_terms_list[g]) - 1)
product_description.append(genre_terms_list[g][word_index])
pid += 1
product_descriptions.append(
Product(pid, ' '.join(product_description)))
behavioral_profiles.append(BehavioralProfile(g, product_descriptions))
Yamlator.dump("behavioral_profiles.yaml", behavioral_profiles)
def generate_profile_data(num_customer_profiles,
terms_per_profile,
min_products_purchased_per_customer,
max_products_purchased_per_customer,
num_products_per_behavioral_profile):
"""
Generates customer and behavioral profile data.
"""
#builds term/count map by genre
genre_freq_ass_map=dict()
print "building user product data set"
for f in glob.glob("seed/*.txt"):
genre = f.split("/")[1].split("-")[0]
if genre not in genre_freq_ass_map:
genre_freq_ass_map[genre]=FrequencyAssessor()
with open (f, "r") as current_file:
text = current_file.read()
tokens = nltk.word_tokenize(text)
for t in tokens:
if t not in string.punctuation and len(t) > 2:
if t.endswith("."):
t=t[:len(t)-1]
genre_freq_ass_map[genre].update(t.lower())
#build genre>list of terms freq of occurance weighted.
genre_terms_list=dict()
for g in genre_freq_ass_map.keys():
genre_terms_list[g]=[]
for t in genre_freq_ass_map[g].get_top_terms(max=500):
for i in range(t[1]):
genre_terms_list[g].append(t[0])
#build customer profiles
customer_profiles=[]
for g in genre_terms_list.keys():
cid=0
for c in range(num_customer_profiles):
product_descriptions=[]
pid=0
for p in range(random.randint(min_products_purchased_per_customer,
max_products_purchased_per_customer)):
product_description=[]
for i in range(terms_per_profile):
word_index = random.randint(0,len(genre_terms_list[g])-1)
product_description.append(genre_terms_list[g][word_index])
z=0
while z<4:
#pop in some random words
ii = random.randint(0,len(genre_terms_list.keys())-1)
random_genre = genre_terms_list.keys()[ii]
word_index = random.randint(0,len(genre_terms_list[random_genre])-1)
product_description.append(genre_terms_list[random_genre][word_index])
z+=1
pid+=1
product_descriptions.append(Product(pid, ' '.join(product_description)))
cid+=1
customer_profiles.append(Customer(g, cid, product_descriptions))
Yamlator.dump("customers.yaml", customer_profiles)
#build behavioral_profiles
behavioral_profiles=[]
for g in genre_terms_list.keys():
product_descriptions=[]
pid=0
for p in range(num_products_per_behavioral_profile):
product_description=[]
for i in range(terms_per_profile):
word_index = random.randint(0,len(genre_terms_list[g])-1)
product_description.append(genre_terms_list[g][word_index])
pid+=1
product_descriptions.append(Product(pid, ' '.join(product_description)))
behavioral_profiles.append(BehavioralProfile(g, product_descriptions))
Yamlator.dump("behavioral_profiles.yaml", behavioral_profiles)
