广告特征

src/main/scala/com/aliyun/odps/spark/examples/makedata_ad/makedata_ad_31_originData_20240620.scala

@@ -0,0 +1,388 @@
+package com.aliyun.odps.spark.examples.makedata_ad
+
+import com.alibaba.fastjson.{JSON, JSONObject}
+import com.aliyun.odps.TableSchema
+import com.aliyun.odps.data.Record
+import com.aliyun.odps.spark.examples.myUtils.{MyDateUtils, MyHdfsUtils, ParamUtils, env}
+import examples.extractor.RankExtractorFeature_20240530
+import org.apache.hadoop.io.compress.GzipCodec
+import org.apache.spark.sql.SparkSession
+import org.xm.Similarity
+
+import scala.collection.JavaConversions._
+import scala.collection.mutable.ArrayBuffer
+/*
+   20240608 提取特征
+ */
+
+object makedata_ad_31_originData_20240620 {
+  def main(args: Array[String]): Unit = {
+    val spark = SparkSession
+      .builder()
+      .appName(this.getClass.getName)
+      .getOrCreate()
+    val sc = spark.sparkContext
+
+    // 1 读取参数
+    val param = ParamUtils.parseArgs(args)
+    val tablePart = param.getOrElse("tablePart", "64").toInt
+    val beginStr = param.getOrElse("beginStr", "2024062008")
+    val endStr = param.getOrElse("endStr", "2024062023")
+    val savePath = param.getOrElse("savePath", "/dw/recommend/model/31_ad_sample_data/")
+    val project = param.getOrElse("project", "loghubods")
+    val table = param.getOrElse("table", "alg_recsys_ad_sample_all")
+    val repartition = param.getOrElse("repartition", "100").toInt
+
+    // 2 读取odps+表信息
+    val odpsOps = env.getODPS(sc)
+
+    // 3 循环执行数据生产
+    val timeRange = MyDateUtils.getDateHourRange(beginStr, endStr)
+    for (dt_hh <- timeRange) {
+      val dt = dt_hh.substring(0, 8)
+      val hh = dt_hh.substring(8, 10)
+      val partition = s"dt=$dt,hh=$hh"
+      println("开始执行partiton:" + partition)
+      val odpsData = odpsOps.readTable(project = project,
+        table = table,
+        partition = partition,
+        transfer = func,
+        numPartition = tablePart)
+        .map(record => {
+
+
+          val ts = record.getString("ts").toInt
+          val cid = record.getString("cid")
+
+
+          val featureMap = new JSONObject()
+
+          val b1: JSONObject = if (record.isNull("b1_feature")) new JSONObject() else
+            JSON.parseObject(record.getString("b1_feature"))
+          val b2: JSONObject = if (record.isNull("b2_feature")) new JSONObject() else
+            JSON.parseObject(record.getString("b2_feature"))
+          val b3: JSONObject = if (record.isNull("b3_feature")) new JSONObject() else
+            JSON.parseObject(record.getString("b3_feature"))
+          val b4: JSONObject = if (record.isNull("b4_feature")) new JSONObject() else
+            JSON.parseObject(record.getString("b4_feature"))
+          val b5: JSONObject = if (record.isNull("b5_feature")) new JSONObject() else
+            JSON.parseObject(record.getString("b5_feature"))
+          val b6: JSONObject = if (record.isNull("b6_feature")) new JSONObject() else
+            JSON.parseObject(record.getString("b6_feature"))
+          val b7: JSONObject = if (record.isNull("b7_feature")) new JSONObject() else
+            JSON.parseObject(record.getString("b7_feature"))
+          val b8: JSONObject = if (record.isNull("b8_feature")) new JSONObject() else
+            JSON.parseObject(record.getString("b8_feature"))
+          val b9: JSONObject = if (record.isNull("b9_feature")) new JSONObject() else
+            JSON.parseObject(record.getString("b9_feature"))
+
+
+          featureMap.put("cid_" + cid, 1.0)
+          if (b1.containsKey("adid") && b1.getString("adid").nonEmpty) {
+            featureMap.put("adid_" + b1.getString("adid"), 1.0)
+          }
+          if (b1.containsKey("adverid") && b1.getString("adverid").nonEmpty) {
+            featureMap.put("adverid_" + b1.getString("adverid"), 1.0)
+          }
+          if (b1.containsKey("targeting_conversion") && b1.getString("targeting_conversion").nonEmpty) {
+            featureMap.put("targeting_conversion_" + b1.getString("targeting_conversion"), 1.0)
+          }
+
+
+          if (b1.containsKey("cpa")) {
+            featureMap.put("cpa", b1.getString("cpa").toDouble)
+          }
+
+          for ((bn, prefix1) <- List(
+            (b2, "b2"), (b3, "b3"),(b4, "b4"),(b5, "b5"),(b8, "b8")
+          )){
+            for (prefix2 <- List(
+              "3h", "6h", "12h", "1d", "3d", "7d"
+            )){
+              val view = if (bn.isEmpty) 0D else bn.getIntValue("ad_view_" + prefix2).toDouble
+              val click = if (bn.isEmpty) 0D else bn.getIntValue("ad_click_" + prefix2).toDouble
+              val conver = if (bn.isEmpty) 0D else bn.getIntValue("ad_conversion_" + prefix2).toDouble
+              val income = if (bn.isEmpty) 0D else bn.getIntValue("ad_income_" + prefix2).toDouble
+              val f1 = RankExtractorFeature_20240530.calDiv(click, view)
+              val f2 = RankExtractorFeature_20240530.calDiv(conver, view)
+              val f3 = RankExtractorFeature_20240530.calDiv(conver, click)
+              val f4 = conver
+              val f5 = RankExtractorFeature_20240530.calDiv(income*1000, view)
+              featureMap.put(prefix1 + "_" + prefix2 + "_" + "ctr", f1)
+              featureMap.put(prefix1 + "_" + prefix2 + "_" + "ctcvr", f2)
+              featureMap.put(prefix1 + "_" + prefix2 + "_" + "cvr", f3)
+              featureMap.put(prefix1 + "_" + prefix2 + "_" + "conver", f4)
+              featureMap.put(prefix1 + "_" + prefix2 + "_" + "ecpm", f5)
+            }
+          }
+
+          for ((bn, prefix1) <- List(
+            (b6, "b6"), (b7, "b7")
+          )) {
+            for (prefix2 <- List(
+              "7d", "14d"
+            )) {
+              val view = if (bn.isEmpty) 0D else bn.getIntValue("ad_view_" + prefix2).toDouble
+              val click = if (bn.isEmpty) 0D else bn.getIntValue("ad_click_" + prefix2).toDouble
+              val conver = if (bn.isEmpty) 0D else bn.getIntValue("ad_conversion_" + prefix2).toDouble
+              val income = if (bn.isEmpty) 0D else bn.getIntValue("ad_income_" + prefix2).toDouble
+              val f1 = RankExtractorFeature_20240530.calDiv(click, view)
+              val f2 = RankExtractorFeature_20240530.calDiv(conver, view)
+              val f3 = RankExtractorFeature_20240530.calDiv(conver, click)
+              val f4 = conver
+              val f5 = RankExtractorFeature_20240530.calDiv(income * 1000, view)
+              featureMap.put(prefix1 + "_" + prefix2 + "_" + "ctr", f1)
+              featureMap.put(prefix1 + "_" + prefix2 + "_" + "ctcvr", f2)
+              featureMap.put(prefix1 + "_" + prefix2 + "_" + "cvr", f3)
+              featureMap.put(prefix1 + "_" + prefix2 + "_" + "conver", f4)
+              featureMap.put(prefix1 + "_" + prefix2 + "_" + "ecpm", f5)
+            }
+          }
+
+          val c1: JSONObject = if (record.isNull("c1_feature")) new JSONObject() else
+            JSON.parseObject(record.getString("c1_feature"))
+
+          val midActionList = if (c1.containsKey("action") && c1.getString("action").nonEmpty){
+            c1.getString("action").split(",").map(r=>{
+              val rList = r.split(":")
+              (rList(0), (rList(1).toInt, rList(2).toInt, rList(3).toInt, rList(4).toInt, rList(5)))
+            }).sortBy(-_._2._1).toList
+          }else {
+            new ArrayBuffer[(String, (Int, Int, Int, Int, String))]().toList
+          }
+          // u特征
+          val viewAll = midActionList.size.toDouble
+          val clickAll = midActionList.map(_._2._2).sum.toDouble
+          val converAll = midActionList.map(_._2._3).sum.toDouble
+          val incomeAll = midActionList.map(_._2._4).sum.toDouble
+          featureMap.put("viewAll", viewAll)
+          featureMap.put("clickAll", clickAll)
+          featureMap.put("converAll", converAll)
+          featureMap.put("incomeAll", incomeAll)
+          featureMap.put("ctr_all", RankExtractorFeature_20240530.calDiv(clickAll, viewAll))
+          featureMap.put("ctcvr_all", RankExtractorFeature_20240530.calDiv(converAll, viewAll))
+          featureMap.put("cvr_all", RankExtractorFeature_20240530.calDiv(clickAll, converAll))
+          featureMap.put("ecpm_all", RankExtractorFeature_20240530.calDiv(incomeAll * 1000, viewAll))
+
+          // ui特征
+          val midTimeDiff = scala.collection.mutable.Map[String, Double]()
+          midActionList.foreach{
+            case (cid, (ts_history, click, conver, income, title)) =>
+              if (!midTimeDiff.contains("timediff_view_" + cid)){
+                midTimeDiff.put("timediff_view_" + cid, 1.0 / ((ts - ts_history).toDouble/3600.0/24.0))
+              }
+              if (!midTimeDiff.contains("timediff_click_" + cid) && click > 0) {
+                midTimeDiff.put("timediff_click_" + cid, 1.0 / ((ts - ts_history).toDouble / 3600.0 / 24.0))
+              }
+              if (!midTimeDiff.contains("timediff_conver_" + cid) && conver > 0) {
+                midTimeDiff.put("timediff_conver_" + cid, 1.0 / ((ts - ts_history).toDouble / 3600.0 / 24.0))
+              }
+          }
+
+          val midActionStatic = scala.collection.mutable.Map[String, Double]()
+          midActionList.foreach {
+            case (cid, (ts_history, click, conver, income, title)) =>
+              midActionStatic.put("actionstatic_view_" + cid, 1.0 + midActionStatic.getOrDefault("actionstatic_view_" + cid, 0.0))
+              midActionStatic.put("actionstatic_click_" + cid, click + midActionStatic.getOrDefault("actionstatic_click_" + cid, 0.0))
+              midActionStatic.put("actionstatic_conver_" + cid, conver + midActionStatic.getOrDefault("actionstatic_conver_" + cid, 0.0))
+              midActionStatic.put("actionstatic_income_" + cid, income + midActionStatic.getOrDefault("actionstatic_income_" + cid, 0.0))
+          }
+
+          if (midTimeDiff.contains("timediff_view_" + cid)){
+            featureMap.put("timediff_view", midTimeDiff.getOrDefault("timediff_view_" + cid, 0.0))
+          }
+          if (midTimeDiff.contains("timediff_click_" + cid)) {
+            featureMap.put("timediff_click", midTimeDiff.getOrDefault("timediff_click_" + cid, 0.0))
+          }
+          if (midTimeDiff.contains("timediff_conver_" + cid)) {
+            featureMap.put("timediff_conver", midTimeDiff.getOrDefault("timediff_conver_" + cid, 0.0))
+          }
+          if (midActionStatic.contains("actionstatic_view_" + cid)) {
+            featureMap.put("actionstatic_view", midActionStatic.getOrDefault("actionstatic_view_" + cid, 0.0))
+          }
+          if (midActionStatic.contains("actionstatic_click_" + cid)) {
+            featureMap.put("actionstatic_click", midActionStatic.getOrDefault("actionstatic_click_" + cid, 0.0))
+          }
+          if (midActionStatic.contains("actionstatic_conver_" + cid)) {
+            featureMap.put("actionstatic_conver", midActionStatic.getOrDefault("actionstatic_conver_" + cid, 0.0))
+          }
+          if (midActionStatic.contains("actionstatic_income_" + cid)) {
+            featureMap.put("actionstatic_income", midActionStatic.getOrDefault("actionstatic_income_" + cid, 0.0))
+          }
+          if (midActionStatic.contains("actionstatic_view_" + cid) && midActionStatic.contains("actionstatic_click_" + cid)) {
+            featureMap.put("actionstatic_ctr", RankExtractorFeature_20240530.calDiv(
+              midActionStatic.getOrDefault("actionstatic_click_" + cid, 0.0),
+              midActionStatic.getOrDefault("actionstatic_view_" + cid, 0.0)
+            ))
+          }
+          if (midActionStatic.contains("actionstatic_view_" + cid) && midActionStatic.contains("timediff_conver_" + cid)) {
+            featureMap.put("actionstatic_ctcvr", RankExtractorFeature_20240530.calDiv(
+              midActionStatic.getOrDefault("timediff_conver_" + cid, 0.0),
+              midActionStatic.getOrDefault("actionstatic_view_" + cid, 0.0)
+            ))
+          }
+          if (midActionStatic.contains("actionstatic_conver_" + cid) && midActionStatic.contains("actionstatic_click_" + cid)) {
+            featureMap.put("actionstatic_cvr", RankExtractorFeature_20240530.calDiv(
+              midActionStatic.getOrDefault("actionstatic_click_" + cid, 0.0),
+              midActionStatic.getOrDefault("timediff_conver_" + cid, 0.0)
+            ))
+          }
+
+          val e1: JSONObject = if (record.isNull("e1_feature")) new JSONObject() else
+            JSON.parseObject(record.getString("e1_feature"))
+          val e2: JSONObject = if (record.isNull("e2_feature")) new JSONObject() else
+            JSON.parseObject(record.getString("e2_feature"))
+          val title = b1.getOrDefault("cidtitle", "").toString
+          if (title.nonEmpty){
+            for ((en, prefix1) <- List((e1, "e1"), (e2, "e2"))){
+              for (prefix2 <- List("tags_3d", "tags_7d", "tags_14d")){
+                if (en.nonEmpty && en.containsKey(prefix2) && en.getString(prefix2).nonEmpty) {
+                  val (f1, f2, f3, f4) = funcC34567ForTags(en.getString(prefix2), title)
+                  featureMap.put(prefix1 + "_" + prefix2 + "_matchnum", f1)
+                  featureMap.put(prefix1 + "_" + prefix2 + "_maxscore", f3)
+                  featureMap.put(prefix1 + "_" + prefix2 + "_avgscore", f4)
+
+                }
+              }
+            }
+          }
+
+          val d1: JSONObject = if (record.isNull("d1_feature")) new JSONObject() else
+            JSON.parseObject(record.getString("d1_feature"))
+          val d2: JSONObject = if (record.isNull("d2_feature")) new JSONObject() else
+            JSON.parseObject(record.getString("d2_feature"))
+
+          if (d1.nonEmpty){
+            for (prefix <- List("3h", "6h", "12h", "1d", "3d", "7d")) {
+              val view = if (!d1.containsKey("ad_view_" + prefix)) 0D else d1.getIntValue("ad_view_" + prefix).toDouble
+              val click = if (!d1.containsKey("ad_click_" + prefix)) 0D else d1.getIntValue("ad_click_" + prefix).toDouble
+              val conver = if (!d1.containsKey("ad_conversion_" + prefix)) 0D else d1.getIntValue("ad_conversion_" + prefix).toDouble
+              val income = if (!d1.containsKey("ad_income_" + prefix)) 0D else d1.getIntValue("ad_income_" + prefix).toDouble
+              val f1 = RankExtractorFeature_20240530.calDiv(click, view)
+              val f2 = RankExtractorFeature_20240530.calDiv(conver, view)
+              val f3 = RankExtractorFeature_20240530.calDiv(conver, click)
+              val f4 = conver
+              val f5 = RankExtractorFeature_20240530.calDiv(income * 1000, view)
+              featureMap.put("d1_feature" + "_" + prefix + "_" + "ctr", f1)
+              featureMap.put("d1_feature" + "_" + prefix + "_" + "ctcvr", f2)
+              featureMap.put("d1_feature" + "_" + prefix + "_" + "cvr", f3)
+              featureMap.put("d1_feature" + "_" + prefix + "_" + "conver", f4)
+              featureMap.put("d1_feature" + "_" + prefix + "_" + "ecpm", f5)
+            }
+          }
+
+          val vidRankMaps = scala.collection.mutable.Map[String, scala.collection.immutable.Map[String, Double]]()
+          if (d2.nonEmpty){
+            d2.foreach(r => {
+              val key = r._1
+              val value = d2.getString(key).split(",").map(r=> {
+                val rList = r.split(":")
+                (rList(0), rList(2).toDouble)
+              }).toMap
+              vidRankMaps.put(key, value)
+            })
+          }
+          for (prefix1 <- List("ctr", "ctcvr", "ecpm")) {
+            for (prefix2 <- List("1d", "3d", "7d", "14d")) {
+              if (vidRankMaps.contains(prefix1 + "_" + prefix2)){
+                val rank = vidRankMaps(prefix1 + "_" + prefix2).getOrDefault(cid, 0.0)
+                if (rank >= 1.0){
+                  featureMap.put("vid_rank_" + prefix1 + "_" + prefix2, 1.0 / rank)
+                }
+              }
+            }
+          }
+
+
+          /*
+          广告
+            sparse：cid adid adverid targeting_conversion
+
+            cpa --> 1个
+            adverid下的 3h 6h 12h 1d 3d 7d 、 ctr ctcvr cvr conver ecpm  --> 30个
+            cid下的 3h 6h 12h 1d 3d 7d 、 ctr ctcvr cvr ecpm conver --> 30个
+            地理//cid下的 3h 6h 12h 1d 3d 7d 、 ctr ctcvr cvr ecpm conver --> 30个
+            app//cid下的 3h 6h 12h 1d 3d 7d 、 ctr ctcvr cvr ecpm conver --> 30个
+            手机品牌//cid下的 3h 6h 12h 1d 3d 7d 、 ctr ctcvr cvr ecpm conver --> 30个
+            系统 无数据
+            week//cid下的 7d 14d、 ctr ctcvr cvr ecpm conver --> 10个
+            hour//cid下的 7d 14d、 ctr ctcvr cvr ecpm conver --> 10个
+
+          用户
+            用户历史 点击/转化 的title tag；3d 7d 14d； cid的title； 数量/最高分/平均分 --> 18个
+            用户历史 14d 看过/点过/转化次数/income； ctr cvr ctcvr ecpm；  --> 8个
+
+            用户到cid的ui特征 --> 10个
+              1/用户最近看过这个cid的时间间隔
+              1/用户最近点过这个cid的时间间隔
+              1/用户最近转过这个cid的时间间隔
+              用户看过这个cid多少次
+              用户点过这个cid多少次
+              用户转过这个cid多少次
+              用户对这个cid花了多少钱
+              用户对这个cid的ctr ctcvr cvr
+
+          视频
+            title与cid的 sim-score-1/-2 无数据
+            vid//cid下的 3h 6h 12h 1d 3d 7d 、 ctr ctcvr cvr ecpm conver --> 30个
+            vid//cid下的 1d 3d 7d 14d、 ctr ctcvr ecpm 的rank值 倒数 --> 12个
+
+           */
+
+
+
+          //4 处理label信息。
+          val labels = new JSONObject
+          for (labelKey <- List("ad_is_click", "ad_is_conversion")){
+            if (!record.isNull(labelKey)){
+              labels.put(labelKey, record.getString(labelKey))
+            }
+          }
+          //5 处理log key表头。
+          val apptype = record.getString("apptype")
+          val mid = record.getString("mid")
+          val headvideoid = record.getString("headvideoid")
+          val logKey = (apptype, mid, cid, ts, headvideoid).productIterator.mkString(",")
+          val labelKey = labels.toString()
+          val featureKey = featureMap.toString()
+          //6 拼接数据，保存。
+          logKey + "\t" + labelKey + "\t" + featureKey
+        })
+
+      // 4 保存数据到hdfs
+      val savePartition = dt + hh
+      val hdfsPath = savePath + "/" + savePartition
+      if (hdfsPath.nonEmpty && hdfsPath.startsWith("/dw/recommend/model/")){
+        println("删除路径并开始数据写入:" + hdfsPath)
+        MyHdfsUtils.delete_hdfs_path(hdfsPath)
+        odpsData.coalesce(repartition).saveAsTextFile(hdfsPath, classOf[GzipCodec])
+      }else{
+        println("路径不合法，无法写入:" + hdfsPath)
+      }
+    }
+  }
+
+  def func(record: Record, schema: TableSchema): Record = {
+    record
+  }
+
+  def funcC34567ForTags(tags: String, title: String): Tuple4[Double, String, Double, Double] = {
+    // 匹配数量 匹配词 语义最高相似度分 语义平均相似度分
+    val tagsList = tags.split(",")
+    var d1 = 0.0
+    val d2 = new ArrayBuffer[String]()
+    var d3 = 0.0
+    var d4 = 0.0
+    for (tag <- tagsList) {
+      if (title.contains(tag)) {
+        d1 = d1 + 1.0
+        d2.add(tag)
+      }
+      val score = Similarity.conceptSimilarity(tag, title)
+      d3 = if (score > d3) score else d3
+      d4 = d4 + score
+    }
+    d4 = if (tagsList.nonEmpty) d4 / tagsList.size else d4
+    (d1, d2.mkString(","), d3, d4)
+  }
+}

src/main/scala/com/aliyun/odps/spark/examples/临时记录的脚本-广告

@@ -0,0 +1,45 @@
+
+
+
+nohup /opt/apps/SPARK2/spark-2.4.8-hadoop3.2-1.0.8/bin/spark-class2 org.apache.spark.deploy.SparkSubmit \
+--class com.aliyun.odps.spark.examples.makedata.makedata_ad_31_originData_20240620 \
+--master yarn --driver-memory 1G --executor-memory 2G --executor-cores 1 --num-executors 16 \
+./target/spark-examples-1.0.0-SNAPSHOT-shaded.jar \
+tablePart:64 repartition:32 \
+beginStr:2024062008 endStr:2024062008 \
+savePath:/dw/recommend/model/31_ad_sample_data/ \
+table:alg_recsys_ad_sample_all \
+> p31_2024062008.log 2>&1 &
+
+
+nohup /opt/apps/SPARK2/spark-2.4.8-hadoop3.2-1.0.8/bin/spark-class2 org.apache.spark.deploy.SparkSubmit \
+--class com.aliyun.odps.spark.examples.makedata.makedata_14_valueData_20240608 \
+--master yarn --driver-memory 1G --executor-memory 3G --executor-cores 1 --num-executors 32 \
+./target/spark-examples-1.0.0-SNAPSHOT-shaded.jar \
+readPath:/dw/recommend/model/13_sample_data/ \
+savePath:/dw/recommend/model/14_feature_data/ \
+beginStr:20240615 endStr:20240615 repartition:1000 \
+> p14_data_check.log 2>&1 &
+
+
+nohup /opt/apps/SPARK2/spark-2.4.8-hadoop3.2-1.0.8/bin/spark-class2 org.apache.spark.deploy.SparkSubmit \
+--class com.aliyun.odps.spark.examples.makedata.makedata_15_bucket_20240608 \
+--master yarn --driver-memory 16G --executor-memory 1G --executor-cores 1 --num-executors 16 \
+--conf spark.driver.maxResultSize=16G \
+./target/spark-examples-1.0.0-SNAPSHOT-shaded.jar \
+readPath:/dw/recommend/model/14_feature_data/20240606/ fileName:20240606_200_v3 \
+bucketNum:200 sampleRate:0.1 \
+> p15_data2.log 2>&1 &
+
+
+nohup /opt/apps/SPARK2/spark-2.4.8-hadoop3.2-1.0.8/bin/spark-class2 org.apache.spark.deploy.SparkSubmit \
+--class com.aliyun.odps.spark.examples.makedata.makedata_16_bucketData_20240609 \
+--master yarn --driver-memory 2G --executor-memory 4G --executor-cores 1 --num-executors 16 \
+./target/spark-examples-1.0.0-SNAPSHOT-shaded.jar \
+beginStr:20240615 endStr:20240615 repartition:1000 \
+> p16_data.log 2>&1 &
+
+
+/dw/recommend/model/13_sample_data/
+/dw/recommend/model/14_feature_data/
+/dw/recommend/model/16_train_data/