recommend-emr-dataprocess/rec_scripts/rov/train/scripts/update_model.sh

#!/bin/bash

abs_path=$(cd `dirname $0`; pwd)
sample_path="${abs_path}/../data"
model_path="${abs_path}/../model"
predict_tools="/data2/jch/package/alphaFM/bin/fm_predict"
predict_param="-core 16"
update_py="${abs_path}/../src/tools/update2oss.py"
monitor_py="${abs_path}/../src/tools/rec_monitor_push.py"

auc_threshold=0.05
online_model_file=""
push_model_name="推荐模型str+"
if(($#==2))
then
  online_model_file=$1
  push_model_name=$2
else
  exit -1
fi

# model size
model_size=$(wc -l $online_model_file| awk '{print $1}')
if [ $model_size -le 10000 ]
then
  echo "$online_model_file lines <= 10000"
  exit -1
fi

# name & date
model_name=$(echo $online_model_file | awk -F"/" '{print $NF}' | awk -F".txt" '{print $1}')
new_model_date=$(echo $online_model_file | awk -F"/" '{print $(NF-1)}')
old_model_date=$(date -d "$new_model_date -1 day" +"%Y%m%d")

# offline model file
new_off_model_file="${model_path}/${new_model_date}/${model_name}_offline.txt"
old_off_model_file="${model_path}/${old_model_date}/${model_name}_offline.txt"

# eval model
check_file="${sample_path}/${new_model_date}_00063.csv"
if [ -f $check_file ]
then
  eval_file_list="${sample_path}/${new_model_date}_00*.csv"

  # eval new
  new_predict_file="${sample_path}/new_predict.txt"
  echo `date` "cat $eval_file_list | $predict_tools -m $new_off_model_file $predict_param -out $new_predict_file"
  echo
  cat $eval_file_list | $predict_tools -m $new_off_model_file $predict_param -out $new_predict_file &
  wait
  sleep 30s

  # eval old
  old_predict_file="${sample_path}/old_predict.txt"
  echo `date` "cat $eval_file_list | $predict_tools -m $old_off_model_file $predict_param -out $old_predict_file"
  echo
  cat $eval_file_list | $predict_tools -m $old_off_model_file $predict_param -out $old_predict_file &
  wait
  sleep 30s

  # compare auc
  auc_tools="${abs_path}/../src/tools/cal_auc.py"
  new_auc=$(python3 $auc_tools --input $new_predict_file)
  old_auc=$(python3 $auc_tools --input $old_predict_file)
  diff_auc=$(echo $new_auc $old_auc | awk '{print $1-$2}')
  flag=$(echo $diff_auc $auc_threshold | awk '{if($1>-$2&&$1<$2){print 1}else{print 0}}')
  if(($flag==1))
  then
    # 上传模型
    echo `date` "python3.6 $update_py --local $online_model_file"
    python3.6 $update_py --local $online_model_file &
    wait
    # 发送通知
    level=info
    msg=" ${push_model_name}更新完成"
    msg+="\n\t - 新模型AUC: ${new_auc}"
    msg+="\n\t - 老模型AUC: ${old_auc}"
  else
    echo "$diff_auc $auc_threshold"
    # 发送通知
    level=error
    msg=" ${push_model_name}更新失败, ${new_model_date}_auc差异为: ${diff_auc}, 大于$auc_threshold"
    msg+="\n\t - 新模型AUC: ${new_auc}"
    msg+="\n\t - 老模型AUC: ${old_auc}"
  fi
  echo `date` "python3 $monitor_py --level $level --model $push_model_name --msg $msg"
  python3 $monitor_py --level "$level" --model "$push_model_name" --msg "$msg"
fi