recommend-emr-dataprocess/recommend/20_vid_avg_score.sh

#!/bin/sh

# 计算不同VID的平均分

set -x

predict_date=$1
model_name=$2
predict_dim=$3

PROJECT_HOME=/root/zhaohp/recommend-emr-dataprocess/
MODEL_PATH=${PROJECT_HOME}/model/20240805/
PREDICT_PATH=${PROJECT_HOME}/predict/recommend/
TXT_PATH=/mnt/disk1/20240729

HADOOP=/opt/apps/HADOOP-COMMON/hadoop-common-current/bin/hadoop
FM_PREDICT=/root/sunmingze/alphaFM/bin/fm_predict

vids=(22895200 22751457 14146727 22847440 22927926 22858609 22974689 22563167 22959023 22970515 22946931 22994781 20720060 22979110)


restore_score() {
    for(( i = 0; i < ${#vids[@]}; i++)) do
        vid=${vids[i]}
        score_avg=$(awk '{
            score = $2
            new_score = ( 0.1 * score ) / ( 1 - 0.9 * score)
            sum += new_score
            count++
        } END {
            if ( count > 0 ){
                print sum / count
            } else {
                print "NaN"
            }
        }' ${PREDICT_PATH}/${model_name}_${predict_date}_${vid}.txt)
        echo -e "VID: ${vid} 平均分计算结果: ${score_avg} \n\t数据路径: ${PREDICT_PATH}/${model_name}_${predict_date}_${vid}.txt"
    done
}

main() {
    for(( i = 0; i < ${#vids[@]}; i++)) do
        vid=${vids[i]}
        cat ${TXT_PATH}/${predict_date}.txt | \
        awk -v vid="$vid" -F'\t' '{
            if ($2 == vid) {
                split($0, fields, "\t");
                OFS="\t";
                line="";
                for (i=1; i<= length(fields); i++){ 
                    if (i != 2) {
                        line = (line ? line "\t" : "") fields[i];
                    }
                }
                print line
            }
        }' | \
        ${FM_PREDICT} -m ${MODEL_PATH}/${model_name}.txt -dim ${predict_dim} -core 8 -out ${PREDICT_PATH}/${model_name}_${predict_date}_${vid}.txt
        score_avg=`awk '{ sum += $2; count++ } END { if (count > 0) print sum / count }' ${PREDICT_PATH}/${model_name}_${predict_date}_${vid}.txt`
        echo -e "VID: ${vid} 平均分计算结果: ${score_avg} \n\t模型路径: ${MODEL_PATH}/${model_name}.txt \n\t评估数据路径: ${TXT_PATH}/${predict_date}.txt"
    done
}

main


# nohup ./recommend/20_vid_avg_score.sh 20240729 model_recommend_v3_sample_01_20240728 8 > logs/20_vid_model_recommend_v3_20240728.sh 2>&1 &