out_row_count = 0
for user_id in user_dict.keys():
user_features, user_features_header = getUserFeatures(
user_dict[user_id], prefecture_location_dict, model_start_date)
try:
userid_coupon_purchase_detail_list = coupon_purchase_detail_dict[
user_id]
except:
userid_coupon_purchase_detail_list = []
user_hist_features, user_hist_features_header, master_list = getUserHistFeatures(
userid_coupon_purchase_detail_list, coupon_list_feature_dict,
model_start_date, model_start_date)
user_visit_features, user_visit_features_header = getUserVisitFeatures(
user_id, coupon_visit_dict, model_start_date)
for coupon_id in coupon_list_train_dict.keys():
coupon_area_features, coupon_area_features_header = getCouponAreaFeatures(
user_dict[user_id], coupon_area_dict, coupon_id)
coupon_features, coupon_features_header = getCouponFeatures(
coupon_list_train_dict[coupon_id], master_list)
dv_value = getDV(user_id, coupon_id, dv_dict)
if out_row_count == 0:
out_header = [
"USER_ID_hash", "COUPON_ID_hash"
] + user_features_header + user_hist_features_header + user_visit_features_header + coupon_area_features_header + coupon_features_header + [
"DV"
]
out_header_len = len(out_header)
out_file.writerow(out_header)
out_row = [
user_id, coupon_id
] + user_features + user_hist_features + user_visit_features + coupon_area_features + coupon_features + [
dv_value
new_coupon_purchase_detail_userid_list = []
for coupon_purchase_detail in coupon_purchase_detail_userid_list[:
ind]:
if datetime.datetime.strptime(coupon_purchase_detail['I_DATE'],
"%Y-%m-%d %H:%M:%S").date(
) < date_before_purchase_date:
new_coupon_purchase_detail_userid_list.append(
coupon_purchase_detail)
#print new_coupon_purchase_detail_userid_list
user_hist_features, user_hist_features_header, master_list = getUserHistFeatures(
new_coupon_purchase_detail_userid_list,
coupon_list_feature_dict, model_start_date,
date_before_purchase_date)
user_visit_features, user_visit_features_header = getUserVisitFeatures(
user_id, coupon_visit_dict, date_before_purchase_date)
coupon_area_features, coupon_area_features_header = getCouponAreaFeatures(
user_dict[user_id], coupon_area_dict, purchase_coupon)
# getting coupon for dv = 1 #
coupon_features, coupon_features_header = getCouponFeatures(
coupon_list_feature_dict[purchase_coupon], master_list)
dv_value = 1
if out_row_count == 0:
out_header = [
"USER_ID_hash", "COUPON_ID_hash"
] + user_features_header + user_hist_features_header + user_visit_features_header + coupon_area_features_header + coupon_features_header + [
"DV"
]
out_header_len = len(out_header)
out_file.writerow(out_header)
out_row = [
user_id, purchase_coupon
except:
date_before_purchase_date = puchase_date
daywise_coupon_list = daywise_coupon_list_dict[date_before_purchase_date]
# getting user features #
user_features, user_features_header = getUserFeatures(user_dict[user_id], prefecture_location_dict, date_before_purchase_date)
# getting the new history list and getting user_hist features based on it #
new_coupon_purchase_detail_userid_list = []
for coupon_purchase_detail in coupon_purchase_detail_userid_list[:ind]:
if datetime.datetime.strptime(coupon_purchase_detail['I_DATE'], "%Y-%m-%d %H:%M:%S").date() < date_before_purchase_date:
new_coupon_purchase_detail_userid_list.append(coupon_purchase_detail)
#print new_coupon_purchase_detail_userid_list
user_hist_features, user_hist_features_header, master_list = getUserHistFeatures(new_coupon_purchase_detail_userid_list, coupon_list_feature_dict, model_start_date, date_before_purchase_date)
user_visit_features, user_visit_features_header = getUserVisitFeatures(user_id, coupon_visit_dict, date_before_purchase_date)
coupon_area_features, coupon_area_features_header = getCouponAreaFeatures(user_dict[user_id], coupon_area_dict, purchase_coupon)
# getting coupon for dv = 1 #
coupon_features, coupon_features_header = getCouponFeatures(coupon_list_feature_dict[purchase_coupon], master_list)
dv_value = 1
if out_row_count == 0:
out_header = ["USER_ID_hash", "COUPON_ID_hash"]+ user_features_header + user_hist_features_header + user_visit_features_header + coupon_area_features_header + coupon_features_header + ["DV"]
out_header_len = len(out_header)
out_file.writerow( out_header )
out_row = [user_id, purchase_coupon] + user_features + user_hist_features + user_visit_features + coupon_area_features + coupon_features + [dv_value]
assert len(out_row) == out_header_len
out_file.writerow( out_row )
out_row_count += 1
# getting some coupons for dv=0 #
no_of_non_purchased_coupons = 4
coupon_visit_dict = getCouponVisitDict(coupon_visit_feature_file) print "Preparing the data.." user_count = 0 out_row_count = 0 for user_id in user_dict.keys(): user_features, user_features_header = getUserFeatures(user_dict[user_id], prefecture_location_dict, model_start_date) try: userid_coupon_purchase_detail_list = coupon_purchase_detail_dict[user_id] except: userid_coupon_purchase_detail_list = [] user_hist_features, user_hist_features_header, master_list = getUserHistFeatures(userid_coupon_purchase_detail_list, coupon_list_feature_dict, model_start_date, model_start_date) user_visit_features, user_visit_features_header = getUserVisitFeatures(user_id, coupon_visit_dict, model_start_date) for coupon_id in coupon_list_train_dict.keys(): coupon_area_features, coupon_area_features_header = getCouponAreaFeatures(user_dict[user_id], coupon_area_dict, coupon_id) coupon_features, coupon_features_header = getCouponFeatures(coupon_list_train_dict[coupon_id], master_list) dv_value = getDV(user_id, coupon_id, dv_dict) if out_row_count == 0: out_header = ["USER_ID_hash", "COUPON_ID_hash"]+ user_features_header + user_hist_features_header + user_visit_features_header + coupon_area_features_header + coupon_features_header + ["DV"] out_header_len = len(out_header) out_file.writerow( out_header ) out_row = [user_id, coupon_id] + user_features + user_hist_features + user_visit_features + coupon_area_features + coupon_features + [dv_value] assert len(out_row) == out_header_len out_file.writerow( out_row ) out_row_count += 1 #break user_count +=1 #break if user_count % 200 == 0: print "Processed Users : ", user_count