def calculate_fq_subpref_from_usr_file():
usr_home = rd.read_in_user_home_subprefs()
subpref_num_users = rd.read_in_subpref_num_users()
subpref_fq = defaultdict(float)
file_name = "/home/sscepano/D4D res/ORGANIZED/SET3/Distr of Num and Fq of Calls/new results -- check the same/Users_and_their_calling_fq.tsv"
f = open(file_name, 'r')
for line in f:
usr, fq = line.split('\t')
usr = int(usr)
fq = float(fq)
subpref_fq[usr_home[usr]] += fq
for subpref in range(256):
if subpref_num_users[subpref] > 0:
subpref_fq[subpref] = subpref_fq[subpref] / float(subpref_num_users[subpref])
file_name2 = "/home/sscepano/D4D res/ORGANIZED/SET3/Distr of Num and Fq of Calls/subpref/Subpref_avg_calling_fq_check.tsv"
f2 = open(file_name2, 'w')
for subpref in range(256):
if subpref_num_users[subpref] > 0:
f2.write(str(subpref) + '\t' + str(subpref_fq[subpref]) + '\n')
return subpref_fq
#calculate_fq_subpref_from_usr_file()
def calculate_traj_subpref_from_usr_file():
usr_home = rd.read_in_user_home_subprefs()
subpref_num_users = rd.read_in_subpref_num_users()
subpref_rg = defaultdict(float)
file_name = "/home/sscepano/D4D res/ORGANIZED/SET3/Clustering/usr res/usr_traj_length.tsv"
f = open(file_name, 'r')
for line in f:
usr, rg = line.split('\t')
usr = int(usr)
rg = float(rg)
subpref_rg[usr_home[usr]] += rg
for subpref in range(256):
if subpref_num_users[subpref] > 0:
subpref_rg[subpref] = subpref_rg[subpref] / float(subpref_num_users[subpref])
file_name2 = "/home/sscepano/D4D res/allstuff/CLUSTERING/subpref res/subpref_avg_traj.tsv"
f2 = open(file_name2, 'w')
for subpref in range(256):
if subpref_num_users[subpref] > 0:
f2.write(str(subpref) + '\t' + str(subpref_rg[subpref]) + '\n')
return subpref_rg
def data_to_files(data):
num_visited_subprefs_per_subpref = defaultdict(int)
num_visited_subprefs_per_subpref_scaled = defaultdict(int)
usr_home = rd.read_in_user_home_subprefs()
num_usrs = rd.read_in_subpref_num_users()
for usr in range(500001):
num_visited_subprefs_per_subpref[usr_home[usr]] += data[usr].sum()
for subpref in range(256):
if num_usrs[subpref] <> 0:
num_visited_subprefs_per_subpref_scaled[subpref] = num_visited_subprefs_per_subpref[subpref] / num_usrs[subpref]
file1 = "/home/sscepano/D4D res/allstuff/distr of num of visited subprefs/1/usr_num_visited_subprefs.tsv"
file2 = "/home/sscepano/D4D res/allstuff/distr of num of visited subprefs/1/subpref_summed_num_visited_subprefs.tsv"
file3 = "/home/sscepano/D4D res/allstuff/distr of num of visited subprefs/1/subpref_scaled_num_visited_subprefs.tsv"
f1 = open(file1, "w")
f2 = open(file2, "w")
f3 = open(file3, "w")
for usr in range(500001):
f1.write(str(usr) + '\t' + str(n.sum(data[usr])) + '\n')
for subpref in range(256):
if num_visited_subprefs_per_subpref[subpref] <> 0:
f2.write(str(subpref) + '\t' + str(num_visited_subprefs_per_subpref[subpref]) + '\n')
for subpref in range(256):
if num_visited_subprefs_per_subpref_scaled[subpref] <> 0:
f3.write(str(subpref) + '\t' + str(num_visited_subprefs_per_subpref_scaled[subpref]) + '\n')
print file1
print file2
print file3
return
def save_data_to_matrix(home_calls, last_usr_loc_n_dist):
usr_home = rd.read_in_user_home_subprefs()
subpref_avg_fq = rd.read_in_subpref_avg_fq()
subpref_num_usr = rd.read_in_subpref_num_users()
subpref_calls = defaultdict(int)
subpref_outside_calls = defaultdict(int)
subpref_pct_inside_calls = defaultdict(int)
subpref_total_traj = defaultdict(int)
file_o = "/home/sscepano/D4D res/allstuff/CLUSTERING/ALL_SUBPREF_clustering_args.tsv"
f_o = open(file_o, "w")
for usr in range(500001):
subpref_calls[usr_home[usr]] += home_calls[usr][0]
subpref_outside_calls[usr_home[usr]] += home_calls[usr][1]
subpref_total_traj[usr_home[usr]] += last_usr_loc_n_dist[usr][1]
print len(subpref_outside_calls)
print len(subpref_calls)
print len(subpref_total_traj)
print len(subpref_avg_fq)
# for usr_id in usr_home:
# if usr_home[usr_id] == 0:
# print usr_id
for subpref_id in range(256):
suma = float(subpref_calls[subpref_id] + subpref_outside_calls[subpref_id])
if suma <> 0:
subpref_pct_inside_calls[subpref_id] = subpref_calls[subpref_id] / suma
subpref_total_traj[subpref_id] = subpref_total_traj[subpref_id] / subpref_num_usr[subpref_id]
f_o.write(str(subpref_id) + '\t' + str(subpref_calls[subpref_id]) + '\t' + str(subpref_outside_calls[subpref_id]) + '\t' + \
str(subpref_pct_inside_calls[subpref_id]) + '\t' + str(subpref_total_traj[subpref_id]) + '\t' + str(subpref_avg_fq[subpref_id]) + '\n')
return
