数据挖掘方面的比赛及代码资料整合(另附NLP相关知识)

关于GBDT原理的相关链接:

XGboost源码解析:https://www.jianshu.com/p/8977a2d97234
XGboost论文翻译:https://blog.csdn.net/qdbszsj/article/details/79615712
GBDT、XGBoost、LightGBM 的使用及参数调优 https://zhuanlan.zhihu.com/p/33700459
GBDT算法原理以及实例理解 https://blog.csdn.net/zpalyq110/article/details/79527653
梯度提升树(GBDT)原理小结 https://www.cnblogs.com/pinard/p/6140514.html
gbdt和xgboost区别 https://blog.csdn.net/zhangf666/article/details/70174464

数据挖掘方面:

场景一:新闻预测股票
NN方案链接:https://www.kaggle.com/arunkumarramanan/market-data-nn-baseline/notebook
EDA方案链接:https://www.kaggle.com/artgor/eda-feature-engineering-and-everything/notebook
方案1链接:https://www.kaggle.com/bguberfain/a-simple-model-using-the-market-and-news-data
方案3链接:https://www.kaggle.com/jsaguiar/baseline-with-news
场景二:销量预测
https://www.kaggle.com/c/rossmann-store-sales
场景三:餐厅选址
数据:https://www.kaggle.com/c/restaurant-revenue-prediction/data
方案:https://www.kaggle.com/jquesadar/1st-place-restaurant-revenue-competition
场景四:用户画像:
比赛链接:https://www.datafountain.cn/competitions/242/details/data-evaluation,
代码链接:https://github.com/feidapeng/2016CCF_StateGrid_UserProfile
场景五:电力预测
比赛链接:https://tianchi.aliyun.com/competition/entrance/231602/information
代码链接:https://github.com/mrzhangboss/electricAI/tree/master/code
场景六:风控
天池平台
https://tianchi.aliyun.com/competition/entrance/231660/introduction?spm=5176.12281957.1004.9.38b04c2aeVsy8R A股上市公司季度营收预测
第3届 融360天机智能金融算法挑战赛
http://openresearch.rong360.com/#/home?source=dc
解决方案代码:https://github.com/Questions1/Rong360
DC数据平台
微额借款用户人品预测大赛
http://www.dcjingsai.com/common/cmpt/微额借款用户人品预测大赛_竞赛信息.html
解决方案代码:https://github.com/Root-lee/ML_micro_loan
用户贷款风险预测
http://www.dcjingsai.com/common/cmpt/用户贷款风险预测_竞赛信息.html
解决方案代码:https://github.com/Questions1/Rong360
信贷违约预测模型(邀请赛)
http://www.dcjingsai.com/common/cmpt/信贷违约预测模型(邀请赛)_赛体与数据.html
解决方案代码:https://github.com/jiguang123/Credit-Loans-of-Data-Analysis
京东金融全球数据探索者大赛:
http://www.dcjingsai.com/common/cmpt/京东金融全球数据探索者大赛_赛体与数据.html
解决方案代码:https://github.com/yaoleiliu/2017-JDD-Global-Data-Explorer-Competition
基于互联网金融的智能信贷风控设计(邀请赛)
http://www.dcjingsai.com/common/cmpt/基于互联网金融的智能信贷风控设计(邀请赛)_竞赛信息.html
datafuntaion平台
消费者人群画像—信用智能评分
https://www.datafountain.cn/competitions/337/details
代码:https://github.com/wangvenn/Credit-Scoring-Regression
海外投资风险智能识别
https://www.datafountain.cn/competitions/250/details
P2P网络借贷平台的经营风险量化分析
https://www.datafountain.cn/competitions/233/details
魔镜杯风控算法大赛
https://ai.ppdai.com/mirror/showCompetitionRisk
代码:https://github.com/wepe/PPD_RiskControlCompetition
风控相关资料:
https://openclub.alipay.com/read.php?tid=7341&fid=96
场景七:供应链预测
比赛链接:https://www.datafountain.cn/competitions/313/details/data-evaluation
代码:https://github.com/yuxiaowww/BDCI-2018-Supply-Chain-Demand-Forecast/blob/master

抽取式自动摘要:

https://blog.csdn.net/qq_32458499/article/details/78659372
https://blog.csdn.net/qq_32458499/article/details/78664199
关于summaRuNNer
https://blog.csdn.net/qq_25222361/article/details/78667850
tensorflow 版实现summaRuNNer
https://github.com/pocheyeniu/SummaRuNNer

实体对齐:

实体对齐综述论文:https://www.ixueshu.com/document/52630e3ac4ffa92d318947a18e7f9386.html
开源框架:https://github.com/wikilinks/nel

实体消歧:

论文导读 | OpenAI的实体消歧新发现 https://zhuanlan.zhihu.com/p/34019499
实体消歧方法分类https://blog.csdn.net/qq_24495287/article/details/87545343
神经网络:https://arxiv.org/abs/1802.01021
github:https://github.com/openai/deeptype

指代消解:

共指消解:
开源工具:https://github.com/huggingface/neuralcoref (该份代码,train里的 conllparser.py,self.extract_mentions缺少方法,当前只能做预测,无法训练自己的模型)
开源工具代码对应的论文:
基础Cluster-ranking模型:https://cs.stanford.edu/people/kevclark/resources/clark-manning-acl16-improving.pdf
强化学习改进:http://www.paperweekly.site/papers/1047
对应源码:https://github.com/clarkkev/deep-coref
注:斯坦福NLP中共指消解使用的神经模型是强化学习版,详情:https://stanfordnlp.github.io/CoreNLP/coref.html
当前 state-of-the-art的论文:https://arxiv.org/pdf/1707.07045.pdf
对应的代码:https://github.com/kentonl/e2e-coref

推荐两个AI网站:
https://paperswithcode.com/sota
https://www.tinymind.cn/

你可能感兴趣的:(数据挖掘比赛)