飞桨学习赛：MarTech Challenge 点击反欺诈预测

Who this competition fits

适合已能使用 PaddlePaddle 2.0+ 训练端到端深度学习模型的人，尤其是想练习广告点击/反欺诈二分类、处理含大量类别字段和时间戳的脱敏表格数据的个人开发者。比赛仅支持个人参赛，参赛者需通过 AI Studio 获取数据、训练资源并提交结果与原始代码。

What this competition gives you

There is a baseline you can run as-is — starting from it is usually far faster than building from scratch.

• DataTraining and test datasets with 500,000 and 150,000 records
• BaselineOfficial baseline system
• TutorialSelected learning project by lastrei
• TutorialSelected learning project by favor
• TutorialSelected learning project by 听闻
• TutorialSelected learning project by GT
• TutorialSelected learning project by FrFFL
• TutorialSelected learning project by weihongwei
• TutorialSelected learning project by 夜光
• TutorialSelected learning project by 炼丹师233
• TutorialSelected learning project by 一霁之寒
• TutorialSelected learning project by jh_crescent
• TutorialSelected learning project by noobimp_
• TutorialSelected learning project by 乡下的老鼠
• TutorialSelected learning project by 风雪夜独酌
• TutorialSelected learning project by 慕容小牛马
• TutorialExcellent project by favor
• TutorialExcellent project by weihongwei
• TutorialExcellent project by cystanford
• TutorialExcellent project by 风离

From registration to a first submission

1. 01

   PaddlePaddle 2.0+

2. 02

   二分类模型训练

3. 03

   表格数据特征工程

4. 04

   类别特征编码

5. 05

   时间戳特征处理

6. 06

   CSV 提交文件生成

Before you commit: 数据中部分特征的物理含义被隐藏且经过脱敏，参赛者需要从包名、设备、网络、地域、时间等字段的统计关联中构造有效信号，而不能依赖完整业务语义。还必须用 PaddlePaddle 2.0+ 生成端到端深度学习模型，并同时提交预测结果和原始代码。