Startsida Utforska Hjälp
Logga in
algorithm
/
recommend-emr-dataprocess
12
0
Fork 0
Filer Ärenden 0 Pull-förfrågningar 1 Wiki
Träd: 661da3da59
Grenar Taggar
recommend-em...
/
zhangbo
/
05_update_everyday_str.sh

05_update_everyday_str.sh 1.9 KB
Historik Rå

123456789101112131415161718192021222324252627282930313233343536373839404142 
  1. #!/bin/sh
    2. 

  2. # 0 全局变量/参数
    3. 

  3. savePath=/dw/recommend/model/share_ratio_samples_v2/
    4. 

  4. model_name=model_sharev2
    5. 

  5. today="$(date -d '1 days ago' +%Y%m%d)"
    6. 

  6. yesterday="$(date -d '2 days ago' +%Y%m%d)"
    7. 

  7. yesterday_early=20231222
    8. 

  8. #yesterday_early="$(date -d '3 days ago' +%Y%m%d)"
    9. 

  9. 

  10. 

  11. HADOOP="/opt/apps/HADOOP-COMMON/hadoop-common-current/bin/hadoop"
    12. 

  12. FM_TRAIN="/root/sunmingze/alphaFM/bin/fm_train"
    13. 

  13. MODEL_PATH="/root/zhangbo/recommend-emr-dataprocess/zhangbo/model/"
    14. 

  14. OSS_PATH="oss://art-recommend.oss-cn-hangzhou.aliyuncs.com/video_str_model"
    15. 

  15. 

  16. # 1 生产数据
    17. 

  17. /opt/apps/SPARK2/spark-2.4.8-hadoop3.2-1.0.8/bin/spark-class2 org.apache.spark.deploy.SparkSubmit \
    18. 

  18. --class com.aliyun.odps.spark.examples.makedata.makedata_01_readtable2hdfs \
    19. 

  19. --master yarn --driver-memory 1G --executor-memory 4G --executor-cores 2 --num-executors 16 \
    20. 

  20. ./target/spark-examples-1.0.0-SNAPSHOT-shaded.jar \
    21. 

  21. tablePart:32 savePath:${savePath} beginStr:${yesterday} endStr:${yesterday}
    22. 

  22. 

  23. # 2 加载上次模型 训练本轮数据 保存本轮模型
    24. 

  24. end_date=${today}
    25. 

  25. loop_date=${yesterday_early}
    26. 

  26. while [[ "$loop_date" != "$end_date" ]]; do
    27. 

  27.     echo -------train ${loop_date}----------
    28. 

  28.     loop_date_model=$(date -d "$loop_date - 1 day" +%Y%m%d)
    29. 

  29.     $HADOOP fs -text ${savePath}/dt=${loop_date}/* | ${FM_TRAIN} -m ${MODEL_PATH}/${model_name}_${loop_date}.txt \
    30. 

  30. -dim 1,1,0 -core 8 -im ${MODEL_PATH}/${model_name}_${loop_date_model}.txt
    31. 

  31.     echo -------save ${MODEL_PATH}/${model_name}_${loop_date}.txt----------
    32. 

  32.     loop_date=$(date -d "$loop_date + 1 day" +%Y%m%d)
    33. 

  33. done
    34. 

  34. 

  35. # 3 本轮模型格式转换
    36. 

  36. cat ${MODEL_PATH}/${model_name}_${yesterday}.txt \
    37. 

  37. | sed '1d' | awk -F " " '{if($2!="0") print $1"\t"$2}' \
    38. 

  38. > ${MODEL_PATH}/${model_name}_${yesterday}_change.txt
    39. 

  39. 

  40. # 4 转换后模型上传oss
    41. 

  41. $HADOOP fs -rm -r oss://art-recommend.oss-cn-hangzhou.aliyuncs.com/video_str_model/${model_name}_change.txt
    42. 

  42. $HADOOP fs -put ${MODEL_PATH}/${model_name}_${yesterday}_change.txt ${OSS_PATH}/${model_name}_change.txt
    43.