Merge branch 'master' of http://git.wanmeizhensuo.com/ML/ffm-baseline

tensnsorflow/multi.py

@@ -41,148 +41,77 @@ def feature_engineer():
     start = (temp - datetime.timedelta(days=2)).strftime("%Y-%m-%d")
     print(start)
 
-    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," \
-          "k.treatment_method,k.price_min,k.price_max,k.treatment_time,k.maintain_time,k.recover_time " \
-          "from jerry_test.esmm_train_data 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 " \
+    sql = "select e.y,e.z,e.stat_date,e.ucity_id,feat.level2_ids " \
+          "from jerry_test.esmm_train_data e " \
           "left join jerry_test.diary_feat feat on e.cid_id = feat.diary_id " \
-          "left join jerry_test.train_Knowledge_network_data k on feat.level2 = k.level2_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"])
-
-    features = ["ucity_id", "ccity_name", "device_type", "manufacturer",
-                "channel", "top", "time", "stat_date", "hospital_id",
-                "treatment_method", "price_min", "price_max", "treatment_time", "maintain_time", "recover_time"]
+    features = ["ucity_id","stat_date"]
 
     df = df.na.fill(dict(zip(features,features)))
 
-    apps_number, app_list_map = multi_hot(df,"app_list",1)
-    level2_number,leve2_map = multi_hot(df,"level2_ids",1 + apps_number)
-    level3_number, leve3_map = multi_hot(df, "level3_ids", 1 + apps_number + level2_number)
+    apps_number, app_list_map = multi_hot(df,"level2_ids",1)
 
     unique_values = []
     for i in features:
         unique_values.extend(df.select(i).distinct().rdd.map(lambda x: x[0]).collect())
-    temp = list(range(2 + apps_number + level2_number + level3_number,
-                      2 + apps_number + level2_number + level3_number + len(unique_values)))
+    temp = list(range(2 + apps_number,
+                      2 + apps_number + len(unique_values)))
     value_map = dict(zip(unique_values, temp))
 
-    rdd = df.select("app_list","level2_ids","level3_ids","stat_date","ucity_id", "ccity_name",
-                    "device_type", "manufacturer", "channel", "top", "time", "hospital_id",
-                    "treatment_method", "price_min","price_max", "treatment_time","maintain_time",
-                    "recover_time","y","z").rdd
+    rdd = df.select("level2_ids","stat_date","ucity_id","y","z").rdd
     rdd.persist()
     # 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.filter(lambda x: x[3]!= validate_date)\
-        .map(lambda x: (app_list_func(x[0], app_list_map), app_list_func(x[1], leve2_map),
-                            app_list_func(x[2], leve3_map),value_map[x[3]],value_map[x[4]],
-                            value_map[x[5]],value_map[x[6]],value_map[x[7]],value_map[x[8]],
-                            value_map[x[9]],value_map[x[10]],value_map[x[11]],value_map[x[12]],
-                            value_map[x[13]],value_map[x[14]],value_map[x[15]],value_map[x[16]],
-                            value_map[x[17]], x[18],x[19]))
-    test = rdd.filter(lambda x: x[3] == validate_date)\
-        .map(lambda x: (app_list_func(x[0], app_list_map), app_list_func(x[1], leve2_map),
-                   app_list_func(x[2], leve3_map), value_map[x[3]], value_map[x[4]],
-                   value_map[x[5]], value_map[x[6]], value_map[x[7]], value_map[x[8]],
-                   value_map[x[9]], value_map[x[10]], value_map[x[11]], value_map[x[12]],
-                   value_map[x[13]], value_map[x[14]], value_map[x[15]], value_map[x[16]],
-                   value_map[x[17]], x[18], x[19]))
-
-    spark.createDataFrame(test).toDF("app_list","level2_ids","level3_ids","stat_date","ucity_id", "ccity_name", "device_type", "manufacturer",
-                  "channel", "top", "time", "hospital_id","treatment_method", "price_min",
-                  "price_max", "treatment_time","maintain_time", "recover_time","y","z")\
-        .repartition(1).write.format("tfrecords").option("recordType", "Example").save(path=path+"va/", mode="overwrite")
+
+    train = rdd.filter(lambda x: x[1]!= validate_date)\
+        .map(lambda x: (app_list_func(x[0], app_list_map),[value_map[x[2]],value_map[x[1]]], x[3],x[4]))
+    test = rdd.filter(lambda x: x[1]== validate_date)\
+        .map(lambda x: (app_list_func(x[0], app_list_map),[value_map[x[2]],value_map[x[1]]], x[3],x[4]))
+
+    spark.createDataFrame(test).toDF("level2_ids","ids","y","z")\
+        .repartition(1).write.format("tfrecords").option("recordType", "SequenceExample").save(path=path+"va/", mode="overwrite")
 
     print("va write done")
-    spark.createDataFrame(train).toDF("app_list","level2_ids","level3_ids","stat_date","ucity_id", "ccity_name", "device_type", "manufacturer",
-                  "channel", "top", "time", "hospital_id","treatment_method", "price_min",
-                  "price_max", "treatment_time","maintain_time", "recover_time","y","z") \
-        .repartition(1).write.format("tfrecords").option("recordType", "Example").save(path=path+"tr/", mode="overwrite")
+    spark.createDataFrame(train).toDF("level2_ids","ids","y","z") \
+        .repartition(1).write.format("tfrecords").option("recordType", "SequenceExample").save(path=path+"tr/", mode="overwrite")
 
     print("done")
     rdd.unpersist()
 
-    return validate_date,value_map,app_list_map,leve2_map,leve3_map
+    return validate_date,value_map,app_list_map
 
 
-def get_predict(date,value_map,app_list_map,level2_map,level3_map):
-    sql = "select e.y,e.z,e.label,e.ucity_id,feat.level2_ids,e.ccity_name," \
-          "u.device_type,u.manufacturer,u.channel,c.top,e.device_id,e.cid_id,cut.time," \
-          "dl.app_list,e.hospital_id,feat.level3_ids," \
-          "k.treatment_method,k.price_min,k.price_max,k.treatment_time,k.maintain_time,k.recover_time " \
-          "from esmm_pre_data e left join user_feature u on e.device_id = u.device_id " \
-          "left join cid_type_top c on e.device_id = c.device_id " \
-          "left join cid_time_cut cut on e.cid_id = cut.cid " \
-          "left join device_app_list dl on e.device_id = dl.device_id " \
-          "left join diary_feat feat on e.cid_id = feat.diary_id " \
-          "left join train_Knowledge_network_data k on feat.level2 = k.level2_id"
+def get_predict(date,value_map,app_list_map):
+    sql = "select e.y,e.z,e.label,e.ucity_id,feat.level2_ids,e.device_id,e.cid_id from esmm_pre_data e " \
+          "left join diary_feat feat on e.cid_id = feat.diary_id limit 50000"
 
-    features = ["app_list", "level2_ids", "level3_ids","ucity_id", "ccity_name", "device_type", "manufacturer",
-                "channel", "top", "time", "hospital_id",
-                "treatment_method", "price_min", "price_max", "treatment_time", "maintain_time", "recover_time"]
+    features = ["ucity_id"]
     df = spark.sql(sql)
-
     df = df.na.fill(dict(zip(features, features)))
-    df = df.drop_duplicates(["ucity_id", "level2_ids", "ccity_name", "device_type", "manufacturer",
-                             "device_id","cid_id","label",
-                             "channel", "top", "time", "app_list", "hospital_id", "level3_ids"])
-
-    rdd = df.select("app_list", "level2_ids", "level3_ids","ucity_id","device_id","cid_id","label", "y", "z",
-                    "ccity_name", "device_type","manufacturer", "channel", "time", "hospital_id",
-              "treatment_method", "price_min", "price_max", "treatment_time", "maintain_time",
-              "recover_time","top") \
-        .rdd.map(lambda x: (app_list_func(x[0], app_list_map), app_list_func(x[1], level2_map),
-                        app_list_func(x[2], level3_map), x[3],x[4],x[5],x[6],x[7],x[8],
-                            value_map.get(x[3], 300000),value_map.get(x[9], 299999),
-                            value_map.get(x[10], 299998), value_map.get(x[11], 299997),
-                            value_map.get(x[12], 299996), value_map.get(x[13], 299995),
-                            value_map.get(x[14], 299994),value_map.get(x[15], 299993),
-                            value_map.get(x[16], 299992),value_map.get(x[17], 299991),
-                            value_map.get(x[18], 299990),value_map.get(x[19], 299989),
-                            value_map.get(x[20], 299988),value_map.get(x[21], 299987),
-                        value_map[date]))
+    rdd = df.select("level2_ids","ucity_id","device_id","cid_id","label", "y", "z") \
+        .rdd.map(lambda x: (app_list_func(x[0], app_list_map),x[1],x[2],x[3],x[4],x[5],x[6],
+                            [value_map.get(x[1], 300000),value_map.get(date, 299999)]))
 
     rdd.persist()
 
-    native_pre = spark.createDataFrame(rdd.filter(lambda x:x[6] == 0).map(lambda x:(x[3],x[4],x[5])))\
+    native_pre = spark.createDataFrame(rdd.filter(lambda x:x[4] == 0).map(lambda x:(x[1],x[2],x[3])))\
         .toDF("city","uid","cid_id")
     print("native")
     native_pre.toPandas().to_csv(local_path+"native.csv", header=True)
 
-    spark.createDataFrame(rdd.filter(lambda x: x[6] == 0)
-                                   .map(lambda x: (x[0], x[1], x[2],x[7],x[8],x[9],x[10],x[11],x[12],
-                                                   x[13],x[14],x[15],
-                                                   x[16],x[17],x[18],x[19],x[20],x[21],x[22],x[23]))) \
-        .toDF("app_list", "level2_ids", "level3_ids","y","z","ucity_id",
-             "ccity_name", "device_type","manufacturer", "channel", "time", "hospital_id",
-              "treatment_method", "price_min", "price_max", "treatment_time", "maintain_time",
-              "recover_time", "top","stat_date").repartition(1).write.format("tfrecords").option("recordType", "Example") \
-        .save(path=path+"native/", mode="overwrite")
-
-    nearby_pre = spark.createDataFrame(rdd.filter(lambda x: x[6] == 1).map(lambda x: (x[3], x[4], x[5]))) \
+    spark.createDataFrame(rdd.filter(lambda x: x[4] == 0).map(lambda x: (x[0],x[5],x[6],x[7]))) \
+        .toDF("level2_ids","y","z","ids").repartition(1).write.format("tfrecords") \
+            .option("recordType", "SequenceExample").save(path=path+"native/", mode="overwrite")
+
+    nearby_pre = spark.createDataFrame(rdd.filter(lambda x: x[4] == 1).map(lambda x:(x[1],x[2],x[3]))) \
         .toDF("city", "uid", "cid_id")
     print("nearby")
     nearby_pre.toPandas().to_csv(local_path+"nearby.csv", header=True)
-
-
-    spark.createDataFrame(rdd.filter(lambda x: x[6] == 1)
-                          .map(lambda x: (x[0], x[1], x[2], x[7], x[8], x[9], x[10], x[11], x[12],
-                                          x[13], x[14], x[15],
-                                          x[16], x[17], x[18], x[19], x[20], x[21], x[22], x[23]))) \
-        .toDF("app_list", "level2_ids", "level3_ids","y","z", "ucity_id",
-               "ccity_name", "device_type", "manufacturer", "channel", "time", "hospital_id",
-              "treatment_method", "price_min", "price_max", "treatment_time", "maintain_time",
-              "recover_time","top","stat_date").repartition(1).write.format("tfrecords").option("recordType", "Example") \
-        .save(path=path+"nearby/", mode="overwrite")
+    spark.createDataFrame(rdd.filter(lambda x: x[4] == 1).map(lambda x: (x[0], x[5], x[6], x[7]))) \
+        .toDF("level2_ids","y","z","ids").repartition(1).write.format("tfrecords") \
+        .option("recordType", "SequenceExample").save(path=path+"nearby/", mode="overwrite")
 
     rdd.unpersist()
 
@@ -196,20 +125,6 @@ def con_sql(db,sql):
     return df
 
 
-def test():
-    sql = "select stat_date,cid_id,y,ccity_name from esmm_train_data limit 60"
-    rdd = spark.sql(sql).select("stat_date","cid_id","y","ccity_name").rdd.map(lambda x:(x[0],x[1],x[2],x[3]))
-    df = spark.createDataFrame(rdd)
-    df.show(6)
-    # spark.sql("use online")
-    # spark.sql("ADD JAR /srv/apps/brickhouse-0.7.1-SNAPSHOT.jar")
-    # spark.sql("ADD JAR /srv/apps/hive-udf-1.0-SNAPSHOT.jar")
-    # spark.sql("CREATE TEMPORARY FUNCTION json_map AS 'brickhouse.udf.json.JsonMapUDF'")
-    # spark.sql("CREATE TEMPORARY FUNCTION is_json AS 'com.gmei.hive.common.udf.UDFJsonFormatCheck'")
-    #
-    # spark.sql("select cl_type from online.tl_hdfs_maidian_view where partition_date = '20190312' limit 6").show()
-
-
 if __name__ == '__main__':
     sparkConf = SparkConf().set("spark.hive.mapred.supports.subdirectories", "true") \
         .set("spark.hadoop.mapreduce.input.fileinputformat.input.dir.recursive", "true") \
@@ -222,13 +137,13 @@ if __name__ == '__main__':
     spark = SparkSession.builder.config(conf=sparkConf).enableHiveSupport().getOrCreate()
     ti = pti.TiContext(spark)
     ti.tidbMapDatabase("jerry_test")
-    ti.tidbMapDatabase("eagle")
+    # ti.tidbMapDatabase("eagle")
     spark.sparkContext.setLogLevel("WARN")
     path = "hdfs:///strategy/esmm/"
-    local_path = "/home/gmuser/test/"
+    local_path = "/home/gmuser/esmm/"
 
-    validate_date, value_map, app_list_map, leve2_map, leve3_map = feature_engineer()
-    get_predict(validate_date, value_map, app_list_map, leve2_map, leve3_map)
+    validate_date, value_map, app_list_map = feature_engineer()
+    get_predict(validate_date, value_map, app_list_map)
 
 
 

tensnsorflow/train_multi.py

@@ -8,10 +8,7 @@
 import shutil
 import os
 import json
-import glob
 from datetime import date, timedelta
-
-import random
 import tensorflow as tf
 
 #################### CMD Arguments ####################
@@ -37,7 +34,8 @@ tf.app.flags.DEFINE_string("deep_layers", '256,128,64', "deep layers")
 tf.app.flags.DEFINE_string("dropout", '0.5,0.5,0.5', "dropout rate")
 tf.app.flags.DEFINE_boolean("batch_norm", False, "perform batch normaization (True or False)")
 tf.app.flags.DEFINE_float("batch_norm_decay", 0.9, "decay for the moving average(recommend trying decay=0.9)")
-tf.app.flags.DEFINE_string("data_dir", '', "data dir")
+tf.app.flags.DEFINE_string("hdfs_dir", '', "hdfs dir")
+tf.app.flags.DEFINE_string("local_dir", '', "local dir")
 tf.app.flags.DEFINE_string("dt_dir", '', "data dt partition")
 tf.app.flags.DEFINE_string("model_dir", '', "model check point dir")
 tf.app.flags.DEFINE_string("servable_model_dir", '', "export servable model for TensorFlow Serving")
@@ -49,19 +47,10 @@ def input_fn(filenames, batch_size=32, num_epochs=1, perform_shuffle=False):
     print('Parsing', filenames)
     def _parse_fn(record):
         features = {
-            "y": tf.FixedLenFeature([], tf.float32),
-            "z": tf.FixedLenFeature([], tf.float32),
-            "ids": tf.FixedLenFeature([FLAGS.field_size], tf.int64),
-            "app_list": tf.VarLenFeature(tf.int64),
-            "level2_list": tf.VarLenFeature(tf.int64),
-            "level3_list": tf.VarLenFeature(tf.int64),
-            "tag1_list": tf.VarLenFeature(tf.int64),
-            "tag2_list": tf.VarLenFeature(tf.int64),
-            "tag3_list": tf.VarLenFeature(tf.int64),
-            "tag4_list": tf.VarLenFeature(tf.int64),
-            "tag5_list": tf.VarLenFeature(tf.int64),
-            "tag6_list": tf.VarLenFeature(tf.int64),
-            "tag7_list": tf.VarLenFeature(tf.int64)
+            "y": tf.VarLenFeature(tf.int64),
+            "z": tf.VarLenFeature(tf.int64),
+            "ids": tf.VarLenFeature(tf.int64),
+            "level2_ids": tf.VarLenFeature(tf.int64)
         }
 
         parsed = tf.parse_single_example(record, features)
@@ -108,15 +97,8 @@ def model_fn(features, labels, mode, params):
 
     feat_ids = features['ids']
     app_list = features['app_list']
-    level2_list = features['level2_list']
-    level3_list = features['level3_list']
-    tag1_list = features['tag1_list']
-    tag2_list = features['tag2_list']
-    tag3_list = features['tag3_list']
-    tag4_list = features['tag4_list']
-    tag5_list = features['tag5_list']
-    tag6_list = features['tag6_list']
-    tag7_list = features['tag7_list']
+    level2_list = features['level2_ids']
+
 
     if FLAGS.task_type != "infer":
         y = labels['y']
@@ -127,18 +109,10 @@ def model_fn(features, labels, mode, params):
         embedding_id = tf.nn.embedding_lookup(Feat_Emb,feat_ids)
         app_id = tf.nn.embedding_lookup_sparse(Feat_Emb, sp_ids=app_list, sp_weights=None, combiner="sum")
         level2 = tf.nn.embedding_lookup_sparse(Feat_Emb, sp_ids=level2_list, sp_weights=None, combiner="sum")
-        level3 = tf.nn.embedding_lookup_sparse(Feat_Emb, sp_ids=level3_list, sp_weights=None, combiner="sum")
-        tag1 = tf.nn.embedding_lookup_sparse(Feat_Emb, sp_ids=tag1_list, sp_weights=None, combiner="sum")
-        tag2 = tf.nn.embedding_lookup_sparse(Feat_Emb, sp_ids=tag2_list, sp_weights=None, combiner="sum")
-        tag3 = tf.nn.embedding_lookup_sparse(Feat_Emb, sp_ids=tag3_list, sp_weights=None, combiner="sum")
-        tag4 = tf.nn.embedding_lookup_sparse(Feat_Emb, sp_ids=tag4_list, sp_weights=None, combiner="sum")
-        tag5 = tf.nn.embedding_lookup_sparse(Feat_Emb, sp_ids=tag5_list, sp_weights=None, combiner="sum")
-        tag6 = tf.nn.embedding_lookup_sparse(Feat_Emb, sp_ids=tag6_list, sp_weights=None, combiner="sum")
-        tag7 = tf.nn.embedding_lookup_sparse(Feat_Emb, sp_ids=tag7_list, sp_weights=None, combiner="sum")
+
 
         # x_concat = tf.reshape(embedding_id,shape=[-1, common_dims])  # None * (F * K)
-        x_concat = tf.concat([tf.reshape(embedding_id,shape=[-1,common_dims]),app_id,level2,level3,tag1,
-                              tag2,tag3,tag4,tag5,tag6,tag7], axis=1)
+        x_concat = tf.concat([tf.reshape(embedding_id,shape=[-1,common_dims]),app_id,level2], axis=1)
 
     with tf.name_scope("CVR_Task"):
         if mode == tf.estimator.ModeKeys.TRAIN:
@@ -301,7 +275,8 @@ def main(_):
 
     print('task_type ', FLAGS.task_type)
     print('model_dir ', FLAGS.model_dir)
-    print('data_dir ', FLAGS.data_dir)
+    print('hdfs_dir ', FLAGS.hdfs_dir)
+    print('local_dir ', FLAGS.local_dir)
     print('dt_dir ', FLAGS.dt_dir)
     print('num_epochs ', FLAGS.num_epochs)
     print('feature_size ', FLAGS.feature_size)
@@ -320,6 +295,7 @@ def main(_):
     path = "hdfs:///strategy/esmm/"
     tr_files = [path+"tr/part-r-00000"]
     va_files = [path+"va/part-r-00000"]
+    te_files = ["%s/part-r-00000" % FLAGS.hdfs_dir]
 
     # tr_files = glob.glob("%s/tr/*tfrecord" % FLAGS.data_dir)
     # random.shuffle(tr_files)
@@ -366,9 +342,9 @@ def main(_):
             print('%s: %s' % (key,value))
     elif FLAGS.task_type == 'infer':
         preds = Estimator.predict(input_fn=lambda: input_fn(te_files, num_epochs=1, batch_size=FLAGS.batch_size), predict_keys=["pctcvr","pctr","pcvr"])
-        with open(FLAGS.data_dir+"/pred.txt", "w") as fo:
+        with open(FLAGS.local_dir+"/pred.txt", "w") as fo:
             print("-"*100)
-            with open(FLAGS.data_dir + "/pred.txt", "w") as fo:
+            with open(FLAGS.local_dir + "/pred.txt", "w") as fo:
                 for prob in preds:
                     fo.write("%f\t%f\t%f\n" % (prob['pctr'], prob['pcvr'], prob['pctcvr']))
     elif FLAGS.task_type == 'export':