recommend-emr-dataprocess/zhangbo/07_ad_model_update_everyday.sh

#!/bin/sh
set -ex

# 0 全局变量/参数
originDataSavePath=/dw/recommend/model/31_ad_sample_data_auto/
bucketFeatureSavePath=/dw/recommend/model/33_ad_train_data_nosparse_auto/
model_name=model_lr0
today="$(date +%Y%m%d)"
today_early_1="$(date -d '1 days ago' +%Y%m%d)"
beginTime=08
endTime=23
beginStr=${today_early_1}${beginTime}
endStr=${today_early_1}${endTime}

MODEL_PATH=/root/zhaohp/recommend-emr-dataprocess/model
HADOOP=/opt/apps/HADOOP-COMMON/hadoop-common-current/bin/hadoop
FM_HOME=/root/sunmingze/alphaFM
OSS_PATH=oss://art-recommend.oss-cn-hangzhou.aliyuncs.com/ad_model/

# 1 判断依赖的数据表是否生产完成
source /root/anaconda3/bin/activate py37
max_hour=15
max_minute=00
while true; do
  python_return_code=$(python zhangbo/utils.py --excute_program check_ad_origin_hive --partition ${today_early_1} --hh ${endTime})
  if [ $python_return_code -eq 0 ]; then
    echo "Python程序返回0，退出循环。"
    break
  fi
  echo "Python程序返回非0值，等待五分钟后再次调用。"
  sleep 300
  current_hour=$(date +%H)
  current_minute=$(date +%M)
  if (( current_hour > max_hour || (current_hour == max_hour && current_minute >= max_minute) )); then
    echo "最长等待时间已到，失败:${current_hour}-${current_minute}"
    exit 1
  fi
done



# 2 原始特征生成
/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.zhp.makedata_ad.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:${beginStr} endStr:${endStr} \
savePath:${originDataSavePath} \
table:alg_recsys_ad_sample_all_new
if [ $? -ne 0 ]; then
   echo "Spark原始样本生产任务执行失败"
   exit 1
else
    echo "spark原始样本生产执行成功"
fi


# 3 特征分桶
/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.zhp.makedata_ad.makedata_ad_33_bucketData_20240622 \
--master yarn --driver-memory 2G --executor-memory 4G --executor-cores 1 --num-executors 16 \
./target/spark-examples-1.0.0-SNAPSHOT-shaded.jar \
beginStr:${today_early_1} endStr:${today_early_1} repartition:400 \
filterNames:XXXXX \
bucketFileName:20240620_ad_bucket_249_fix.txt \
readPath:${originDataSavePath} \
savePath:${bucketFeatureSavePath}
if [ $? -ne 0 ]; then
   echo "Spark特征分桶处理任务执行失败"
   exit 1
else
   echo "spark特征分桶处理执行成功"
fi


# 4 模型训练
$HADOOP fs -text ${bucketFeatureSavePath}/${today_early_1}/* | /root/sunmingze/alphaFM/bin/fm_train -m model/${model_name}_${today_early_1}.txt -dim 1,1,0 -core 8
if [ $? -ne 0 ]; then
   echo "模型训练失败"
   exit 1
fi


# 5 对比AUC

# 6 模型格式转换
cat ${MODEL_PATH}/${model_name}_${today_early_1}.txt \
| sed '1d' | awk -F " " '{if($2!="0") print $1"\t"$2}' \
> ${MODEL_PATH}/${model_name}_${today_early_1}_change.txt

# 7 模型文件上传OSS