# Alibaba_AI_Ops_Competition_Rank14 **Repository Path**: wwdguu/Alibaba_AI_Ops_Competition_Rank14 ## Basic Information - **Project Name**: Alibaba_AI_Ops_Competition_Rank14 - **Description**: 阿里智能运维算法大赛 复赛14名方案 - **Primary Language**: Python - **License**: Not specified - **Default Branch**: master - **Homepage**: None - **GVP Project**: No ## Statistics - **Stars**: 0 - **Forks**: 0 - **Created**: 2020-09-14 - **Last Updated**: 2020-12-19 ## Categories & Tags **Categories**: Uncategorized **Tags**: None ## README # [天池智能运维算法大赛](https://tianchi.aliyun.com/competition/entrance/231775/introduction)-fengyang95-Rank14 [天池智能运维大赛2020踩坑分享](https://tianchi.aliyun.com/forum/postDetail?spm=5176.12586969.1002.15.3f9a77c3U6M5tP&postId=102983) [答辩PPT](https://github.com/fengyang95/Alibaba_AI_Ops_Competition_Rank14/blob/master/Alibaba_AIOps_fengyang95_%E6%9D%8E%E5%85%83%E9%B9%8F.pdf) ## 解决方案 1. 利用pyspark处理原始log数据,打标签 2. 特征包括 原始特征+窗口特征+组合特征 3. 利用self-paced ensemble+lightgbm 训练和测试 ## 代码运行说明 1. 运行 model文件夹内train_pandas_cls_window_7_spe_66_lgb_F37.py 处理数据和训练模型(原始数据解压后放在data/round1_train/logs文件夹) 2. 运行 code文件夹内main.py文件预测结果(线上分数最高的那个模型路径为user_data/model_data/spelgb_window7_66_leaves50_F37)