Skip to content

F
ffm-baseline
Commit 08d18203 authored by 张彦钊's avatar 张彦钊
Browse files
Options

change test file

parent 35ee7eb2
Hide whitespace changes
Inline Side-by-side
Showing with 71 additions and 66 deletions
tensnsorflow/feature_test.py
View file @ 08d18203
......@@ -173,72 +173,72 @@ def feature_engineer():
16 + apps_number + level2_number + level3_number + len(unique_values)))
value_map = dict(zip(unique_values, temp))
# sql = "select e.y,e.z,e.stat_date,e.ucity_id,feat.level2_ids,e.ccity_name,u.device_type,u.manufacturer," \
# "u.channel,c.top,cut.time,dl.app_list,feat.level3_ids,doctor.hospital_id," \
# "wiki.tag as tag1,question.tag as tag2,search.tag as tag3,budan.tag as tag4," \
# "ot.tag as tag5,sixin.tag as tag6,cart.tag as tag7," \
# "k.treatment_method,k.price_min,k.price_max,k.treatment_time,k.maintain_time,k.recover_time " \
# "from jerry_test.esmm_train_data_dwell e left join jerry_test.user_feature u on e.device_id = u.device_id " \
# "left join jerry_test.cid_type_top c on e.device_id = c.device_id " \
# "left join jerry_test.cid_time_cut cut on e.cid_id = cut.cid " \
# "left join jerry_test.device_app_list dl on e.device_id = dl.device_id " \
# "left join jerry_test.diary_feat feat on e.cid_id = feat.diary_id " \
# "left join jerry_test.knowledge k on feat.level2 = k.level2_id " \
# "left join jerry_test.wiki_tag wiki on e.device_id = wiki.device_id " \
# "left join jerry_test.question_tag question on e.device_id = question.device_id " \
# "left join jerry_test.search_tag search on e.device_id = search.device_id " \
# "left join jerry_test.budan_tag budan on e.device_id = budan.device_id " \
# "left join jerry_test.order_tag ot on e.device_id = ot.device_id " \
# "left join jerry_test.sixin_tag sixin on e.device_id = sixin.device_id " \
# "left join jerry_test.cart_tag cart on e.device_id = cart.device_id " \
# "left join eagle.src_zhengxing_api_service service on e.diary_service_id = service.id " \
# "left join eagle.src_zhengxing_api_doctor doctor on service.doctor_id = doctor.id " \
# "where e.stat_date >= '{}'".format(start)
#
# df = spark.sql(sql)
#
# df = df.drop_duplicates(["ucity_id", "level2_ids", "ccity_name", "device_type", "manufacturer",
# "channel", "top", "time", "stat_date", "app_list", "hospital_id", "level3_ids",
# "tag1", "tag2", "tag3", "tag4", "tag5", "tag6", "tag7"])
#
# df = df.na.fill(dict(zip(features, features)))
#
# rdd = df.select("stat_date", "y", "z", "app_list", "level2_ids", "level3_ids",
# "tag1", "tag2", "tag3", "tag4", "tag5", "tag6", "tag7",
# "ucity_id", "ccity_name", "device_type", "manufacturer", "channel", "top", "time",
# "hospital_id", "treatment_method", "price_min", "price_max", "treatment_time",
# "maintain_time", "recover_time").rdd.repartition(200).map(
# lambda x: (x[0], float(x[1]), float(x[2]), app_list_func(x[3], app_list_map), app_list_func(x[4], leve2_map),
# app_list_func(x[5], leve3_map), app_list_func(x[6], leve2_map), app_list_func(x[7], leve2_map),
# app_list_func(x[8], leve2_map), app_list_func(x[9], leve2_map), app_list_func(x[10], leve2_map),
# app_list_func(x[11], leve2_map), app_list_func(x[12], leve2_map),
# [value_map[x[0]], value_map[x[13]], value_map[x[14]], value_map[x[15]], value_map[x[16]],
# value_map[x[17]], value_map[x[18]], value_map[x[19]], value_map[x[20]], value_map[x[21]],
# value_map[x[22]], value_map[x[23]], value_map[x[24]], value_map[x[25]], value_map[x[26]]]))\
# .zipWithIndex().map(lambda x:(x[0][0],x[0][1],x[0][2],x[0][3],x[0][4],x[0][5],x[0][6],x[0][7],x[0][8],
# x[0][9],x[0][10],x[0][11],x[0][12],x[0][13],
# x[1]))
#
#
# rdd.persist(storageLevel= StorageLevel.MEMORY_ONLY_SER)
#
# # TODO 上线后把下面train fliter 删除，因为最近一天的数据也要作为训练集
#
# train = rdd.map(
# lambda x: (x[1], x[2], x[3], x[4], x[5], x[6], x[7], x[8], x[9],
# x[10], x[11], x[12], x[13],x[14]))
# f = time.time()
# spark.createDataFrame(train).toDF("y", "z", "app_list", "level2_list", "level3_list",
# "tag1_list", "tag2_list", "tag3_list", "tag4_list",
# "tag5_list", "tag6_list", "tag7_list", "ids","number") \
# .repartition(1).write.format("tfrecords").save(path=path + "test_tr/", mode="overwrite")
# h = time.time()
# print("train tfrecord done")
# print((h - f) / 60)
#
# print("训练集样本总量：")
# print(rdd.count())
#
sql = "select e.y,e.z,e.stat_date,e.ucity_id,feat.level2_ids,e.ccity_name,u.device_type,u.manufacturer," \
"u.channel,c.top,cut.time,dl.app_list,feat.level3_ids,doctor.hospital_id," \
"wiki.tag as tag1,question.tag as tag2,search.tag as tag3,budan.tag as tag4," \
"ot.tag as tag5,sixin.tag as tag6,cart.tag as tag7," \
"k.treatment_method,k.price_min,k.price_max,k.treatment_time,k.maintain_time,k.recover_time " \
"from jerry_test.esmm_train_data_dwell e left join jerry_test.user_feature u on e.device_id = u.device_id " \
"left join jerry_test.cid_type_top c on e.device_id = c.device_id " \
"left join jerry_test.cid_time_cut cut on e.cid_id = cut.cid " \
"left join jerry_test.device_app_list dl on e.device_id = dl.device_id " \
"left join jerry_test.diary_feat feat on e.cid_id = feat.diary_id " \
"left join jerry_test.knowledge k on feat.level2 = k.level2_id " \
"left join jerry_test.wiki_tag wiki on e.device_id = wiki.device_id " \
"left join jerry_test.question_tag question on e.device_id = question.device_id " \
"left join jerry_test.search_tag search on e.device_id = search.device_id " \
"left join jerry_test.budan_tag budan on e.device_id = budan.device_id " \
"left join jerry_test.order_tag ot on e.device_id = ot.device_id " \
"left join jerry_test.sixin_tag sixin on e.device_id = sixin.device_id " \
"left join jerry_test.cart_tag cart on e.device_id = cart.device_id " \
"left join eagle.src_zhengxing_api_service service on e.diary_service_id = service.id " \
"left join eagle.src_zhengxing_api_doctor doctor on service.doctor_id = doctor.id " \
"where e.stat_date >= '{}'".format(start)
df = spark.sql(sql)
df = df.drop_duplicates(["ucity_id", "level2_ids", "ccity_name", "device_type", "manufacturer",
"channel", "top", "time", "stat_date", "app_list", "hospital_id", "level3_ids",
"tag1", "tag2", "tag3", "tag4", "tag5", "tag6", "tag7"])
df = df.na.fill(dict(zip(features, features)))
rdd = df.select("stat_date", "y", "z", "app_list", "level2_ids", "level3_ids",
"tag1", "tag2", "tag3", "tag4", "tag5", "tag6", "tag7",
"ucity_id", "ccity_name", "device_type", "manufacturer", "channel", "top", "time",
"hospital_id", "treatment_method", "price_min", "price_max", "treatment_time",
"maintain_time", "recover_time").rdd.repartition(200).map(
lambda x: (x[0], float(x[1]), float(x[2]), app_list_func(x[3], app_list_map), app_list_func(x[4], leve2_map),
app_list_func(x[5], leve3_map), app_list_func(x[6], leve2_map), app_list_func(x[7], leve2_map),
app_list_func(x[8], leve2_map), app_list_func(x[9], leve2_map), app_list_func(x[10], leve2_map),
app_list_func(x[11], leve2_map), app_list_func(x[12], leve2_map),
[value_map[x[0]], value_map[x[13]], value_map[x[14]], value_map[x[15]], value_map[x[16]],
value_map[x[17]], value_map[x[18]], value_map[x[19]], value_map[x[20]], value_map[x[21]],
value_map[x[22]], value_map[x[23]], value_map[x[24]], value_map[x[25]], value_map[x[26]]]))\
.zipWithIndex().map(lambda x:(x[0][0],x[0][1],x[0][2],x[0][3],x[0][4],x[0][5],x[0][6],x[0][7],x[0][8],
x[0][9],x[0][10],x[0][11],x[0][12],x[0][13],
x[1]))
rdd.persist(storageLevel= StorageLevel.MEMORY_ONLY_SER)
# TODO 上线后把下面train fliter 删除，因为最近一天的数据也要作为训练集
train = rdd.map(
lambda x: (x[1], x[2], x[3], x[4], x[5], x[6], x[7], x[8], x[9],
x[10], x[11], x[12], x[13],x[14]))
f = time.time()
spark.createDataFrame(train).toDF("y", "z", "app_list", "level2_list", "level3_list",
"tag1_list", "tag2_list", "tag3_list", "tag4_list",
"tag5_list", "tag6_list", "tag7_list", "ids","number") \
.repartition(1).write.format("tfrecords").save(path=path + "test_tr/", mode="overwrite")
h = time.time()
print("train tfrecord done")
print((h - f) / 60)
print("训练集样本总量：")
print(rdd.count())
# get_pre_number()
#
# test = rdd.filter(lambda x: x[0] == validate_date).map(
......@@ -285,6 +285,9 @@ def get_predict(date,value_map,app_list_map,leve2_map,leve3_map):
df = spark.sql(sql)
df = df.drop_duplicates(["ucity_id", "device_id", "cid_id"])
print("pre")
print(df.count())
df = df.na.fill(dict(zip(features, features)))
f = time.time()
rdd = df.select("label", "y", "z", "ucity_id", "device_id", "cid_id", "app_list", "level2_ids", "level3_ids",
......
tensnsorflow/multi_hot.py
View file @ 08d18203
......@@ -300,6 +300,8 @@ def get_predict(date,value_map,app_list_map,leve2_map,leve3_map):
df = spark.sql(sql)
df = df.drop_duplicates(["ucity_id", "device_id", "cid_id"])
print("pre")
print(df.count())
df = df.na.fill(dict(zip(features, features)))
f = time.time()
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment