"p2w_trans",
"CAFE20_RECENCY_SRKIT",
"max_MC_rev",
"CAFE20_P1Y_VISITS_DAY",
"max_MC_Quantity",
labels,
]
catfeatures = [
"CAFE20_gender", "CAFE20_region", "CAFE20_levels", "is_festival_user",
"level_use", "is_LAST_2YEAR_DD_ACTIVE", "cafe_tag_is_mop_available",
"is_merch_user", "p4week_active", "is_LAST_1YEAR_DD_ACTIVE",
"msr_lifestatus", "IS_SR_KIT_USER", "member_monetary"
]
# 数据预处理
df_train, df_btest = data_clean2(df)
df_train = df_train[select_columns]
df_btest = df_btest[select_columns]
for cats in catfeatures:
df_train[cats] = df_train[cats].astype(int)
df_btest[cats] = df_btest[cats].astype(int)
# # 抽样
# df_train = df_train.sample(n=None, frac=0.1, replace=False, weights=None,
# random_state=0, axis=0)
# df_btest = df_btest.sample(n=None, frac=0.1, replace=False, weights=None,
# random_state=0, axis=0)
print(
'正/负',
'DD_end_gap',
'MC_end_gap',
'p2w_amt',
'cafe_tag_p6m_merch_qty',
'MCoffer_red',
'p2w_trans',
'CAFE20_RECENCY_SRKIT',
labels,
]
catfeatures = [
'is_festival_user', 'is_LAST_2YEAR_DD_ACTIVE',
'cafe_tag_is_mop_available', 'IS_SR_KIT_USER'
]
# 数据预处理
df_train = data_clean2(df)
df_btest = data_clean2(df_btest)
df_train = df_train[select_columns]
df_btest = df_btest[select_columns]
for cats in catfeatures:
df_train[cats] = df_train[cats].astype(int)
df_btest[cats] = df_btest[cats].astype(int)
# # 抽样
# df_train = df_train.sample(n=None, frac=0.1, replace=False, weights=None,
# random_state=0, axis=0)
# df_btest = df_btest.sample(n=None, frac=0.1, replace=False, weights=None,
# random_state=0, axis=0)
print(
"class_rank_fillna",
"student_province_byphone",
"subject_ids",
"student_grade_lpo",
"school_background",
"is_first_trail",
"class_background_label",
'student_grade',
# "exam_year",
"coil_in"
]
# 数据预处理
df_train, df_btest = data_clean2(df,
min_date="2018-08-21",
mid_date="2018-12-03",
max_date="2018-12-18",
label=labels)
for f in cat_features:
df_train[f] = df_train[f].astype(int)
df_btest[f] = df_btest[f].astype(int)
df_train = df_train[select_columns]
df_btest = df_btest[select_columns]
# 抽样
df_train = df_train.sample(n=None,
frac=0.1,
replace=False,
weights=None,
random_state=0,
axis=0)
from data_treatment import data_clean2
from data_treatment import load_data_new
import joblib
import pandas as pd
import matplotlib.pyplot as plt
if __name__ == '__main__':
sql = " "
df = load_data_new(sql, filename="btest.csv")
df = data_clean2(df)
labels = "target_is_DD_ACTIVE"
select_columns = [
'is_festival_user',
'is_LAST_2YEAR_DD_ACTIVE',
'cafe_tag_is_mop_available',
'IS_SR_KIT_USER',
'level_use',
'skr_rate',
'merch_rate',
'active_index',
'cafe_tag_p6m_food_qty',
'DD_rev',
'svc_revenue',
'SR_KIT_NUM',
'cafe_tag_p3m_merch_party_size',
'CAFE20_VISIT_MERCH',
'CAFE20_AMT',
'cafe_tag_p3m_food_qty',
'p3m_weekday_trans',
'max_DD_rev',
'DD_end_gap',
df = load_data_new(sql, filename="df_20190215.csv")
student_sql = """SELECT student_id,count(student_id) from trail_pigeon
GROUP BY student_id"""
student_mul = load_data_new(student_sql, filename="student_mul.csv")
student_ids = student_mul[student_mul["count(student_id)"] == 1]
df = df[df["student_id"].isin(student_mul["student_id"])]
label_by_contract = "is_pigeon"
labels = label_by_contract
# 数据预处理
df = data_clean2(df,
min_date="2018-05-01",
mid_date="2018-09-15",
max_date="2018-09-30",
label=labels)
print("data_count", len(df))
drop_features = [
"order_id", "lesson_plan_id", "student_id", "student_no",
"recent_scores", "order_apply_time", "adjust_start_time"
]
df = df.drop(drop_features, axis=1)
#单因素方差分析
df_anova_single = pd.DataFrame()
i = 0
columns = list(df.columns)