Kaggle-Ensemble-Guide
=====================
A combination of Model Ensembling methods that is extremely useful for increasing accuracy of Kaggle's submission.
For more information: http://mlwave.com/kaggle-ensembling-guide/
## Example:
$ python correlations.py ./samples/method1.csv ./samples/method2.csv
Finding correlation between: ./samples/method1.csv and ./samples/method2.csv
Column to be measured: Label
Pearson's correlation score: 0.67898
Kendall's correlation score: 0.66667
Spearman's correlation score: 0.71053
$ python kaggle_vote.py "./samples/method*.csv" "./samples/kaggle_vote.csv"
parsing: ./samples/method1.csv
parsing: ./samples/method2.csv
parsing: ./samples/method3.csv
wrote to ./samples/kaggle_vote.csv
$ python kaggle_vote.py "./samples/_*.csv" "./samples/kaggle_vote.csv" "weighted"
parsing: ./samples/_w3_method1.csv
Using weight: 3
parsing: ./samples/_w2_method2.csv
Using weight: 2
parsing: ./samples/_w2_method3.csv
Using weight: 2
wrote to ./samples/kaggle_vote.csv
$ python kaggle_rankavg.py "./samples/method*.csv" "./samples/kaggle_rankavg.csv"
parsing: ./samples/method1.csv
parsing: ./samples/method2.csv
parsing: ./samples/method3.csv
wrote to ./samples/kaggle_rankavg.csv
$ python kaggle_avg.py "./samples/method*.csv" "./samples/kaggle_avg.csv"
parsing: ./samples/method1.csv
parsing: ./samples/method2.csv
parsing: ./samples/method3.csv
wrote to ./samples/kaggle_avg.csv
$ python kaggle_geomean.py "./samples/method*.csv" "./samples/kaggle_geomean.csv"
parsing: ./samples/method1.csv
parsing: ./samples/method2.csv
parsing: ./samples/method3.csv
wrote to ./samples/kaggle_geomean.csv
## Result:
==> ./samples/method1.csv <==
ImageId,Label
1,1
2,0
3,9
4,9
5,3
==> ./samples/method2.csv <==
ImageId,Label
1,2
2,0
3,6
4,2
5,3
==> ./samples/method3.csv <==
ImageId,Label
1,2
2,0
3,9
4,2
5,3
==> ./samples/kaggle_avg.csv <==
ImageId,Label
1,1.666667
2,0.000000
3,8.000000
4,4.333333
5,3.000000
==> ./samples/kaggle_rankavg.csv <==
ImageId,Label
1,0.25
2,0.0
3,1.0
4,0.5
5,0.75
==> ./samples/kaggle_vote.csv <==
ImageId,Label
1,2
2,0
3,9
4,2
5,3
==> ./samples/kaggle_geomean.csv <==
ImageId,Label
1,1.587401
2,0.000000
3,7.862224
4,3.301927
5,3.000000
没有合适的资源?快使用搜索试试~ 我知道了~
温馨提示
【资源说明】 1、该资源包括项目的全部源码,下载可以直接使用! 2、本项目适合作为计算机、数学、电子信息等专业的竞赛项目学习资料,作为参考学习借鉴。 3、本资源作为“参考资料”如果需要实现其他功能,需要能看懂代码,并且热爱钻研,自行调试。 ByteCup国际机器学习竞赛源码+学习说明.zip
资源推荐
资源详情
资源评论
收起资源包目录
2016ByteCup国际机器学习竞赛源码+学习说明.zip (72个子文件)
code_20105
data
invited_info_train.txt 15.94MB
question_info.txt 937KB
test_nolabel.txt 1.9MB
validate_nolabel.txt 1.92MB
user_info.txt 4.77MB
solution_1
newfeatures
train.csv 37.72MB
feature_score_1.csv 234B
test.csv 4.62MB
svm_bytecup.py 2KB
ensemble.py 597B
xgb_bytecup.py 5KB
model
xgb.model 1.25MB
solution_3
ItemSimilarityRecommender.ipynb 76KB
train_1115_huidalv_improve.csv 32.18MB
ensemble
blend.py 3KB
0-1.py 486B
labelencoder.ipynb 22KB
fm.ipynb 1.16MB
Kaggle-Ensemble-Guide
kaggle_pbr
load_data.py 836B
blend.py 4KB
README.md 131B
Data
train.csv 17.76MB
test.csv 11.83MB
correlations.py 846B
kaggle_vote.py 2KB
blend_proba.py 4KB
samples
kaggle_avg.csv 69B
kaggle_rankavg.csv 46B
kaggle_geomean.csv 69B
method1.csv 34B
method2.csv 34B
method3.csv 34B
kaggle_vote.csv 0B
kaggle_geomean.py 1KB
kaggle_rankavg.py 1KB
kaggle_avg.py 904B
README.md 2KB
Data
train.csv 17.76MB
test.csv 11.83MB
user_label.csv 8.9MB
feature.sql 30KB
fm_rk.ipynb 880KB
validate_1115_for_predict.csv 4.3MB
test_1115_for_predict.csv 3.93MB
output
output.py 741B
推荐相关资料
Svdfeature-1.2.2.zip 149KB
Rendle2010FM.pdf 187KB
libfm-1.42.manual.pdf 286KB
Svdfeature-manual.pdf 545KB
libfm-1.42.src.tar.gz 518KB
APEX-TR-2011-07-11.pdf 539KB
2016Bytecup_Top10.jpg 233KB
大神分享
第二名解决方案.jpg 1.26MB
第一名解决方案BrickMover_Solution.pdf 473KB
2016 Byte Cup国际机器学习竞赛解决方案.docx 112KB
因果推理建模法.R 3KB
第三名解决方案.jpg 97KB
第一名解决方案.jpg 1.05MB
feature engineering
bytecup_fc.py 6KB
feature.sql 30KB
bytecup_fc.sql 18KB
README.md 1KB
solution_2
itembased_uip.csv 19.34MB
data
sim.csv 1.77MB
ratings.csv 16.47MB
train.csv 15.94MB
test.csv 1.95MB
ItemCF.py 8KB
output_pui.csv 12B
UserCF.py 8KB
userbased_uip.csv 20.59MB
evaluation
ndcg.py 6KB
共 72 条
- 1
资源评论
土豆片片
- 粉丝: 1575
- 资源: 5636
上传资源 快速赚钱
- 我的内容管理 展开
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功