本项目旨在品友互动实时竞价数据形式化为格式以供进一步研究data description描述.pdfVIP

make-ipinyou-data

ThisprojectistoformalisetheiPinYouRTBdataintoastandardformatforfurther

researches.

Step0

Gototodownloadipinyou.contest.dataset.zip.

Unzipitandgetthefolderipinyou.contest.dataset.

Step1

Updatethesoftlinkforthefolderipinyou.contest.datasetinoriginal-data.

weinan@ZHANG:~/Project/make-ipinyou-data/original-data$ln-sfn

~/Data/ipinyou.contest.datasetipinyou.contest.datasetUndermake-ipinyou-

data/original-data/ipinyou.contest.datasetthereshouldbetheoriginaldataset

fileslikethis:weinan@ZHANG:~/Project/make-ipinyou-data/original-

data/ipinyou.contest.dataset$lsalgo.submission.demo.tar.bz2README

testing2ndtraining3rd.txt.txttesting3rd

.txtcity.en.txtregion.en.txttraining1st

file.tags.en.txtfiles.md5testing1sttraining2ndYoudonotneedto

furtherunzipthepackagesinthesubfolders.

Step2

Undermake-ipinyou-datafolder,justrunmakeall.

Note:Theuseragentparserusedinformalizeua.pyusesthepythonpackageuser-

agents.

Aftertheprogramfinished,thetotalsizeofthefolderwillbe14G.Thefilesundermake-

ipinyou-datashouldbelikethis:weinan@ZHANG:~/Project/make-ipinyou-data$ls

14582261299733863476LICENSEmkyzxdata.shpythonschema.txt22592821

33583427allMakefileoriginal-dataREADME.mdNormally,weonlydoexperiment

foreachcampaign(e.g.1458).allisjustthemergeofallthecampaigns.Youcandelete

allifyouthinkitisunusefulinyourexperiment.

Useofthedata

Weusecampaign1458aamplehere.weinan@ZHANG:~/Project/make-ipinyou-

data/1458$lsfeatindex.txttest.log.txttest.yzx.txttrain.log.txt

train.yzx.txt*train.log.txtandtest.log.txtaretheformalisedstringdatafor

eachrow(record)intrainandtest.Thefirstcolumniswhethertheuserclicktheadornot.

The14thcolumnisthewinningpriceforthisauction.*featindex.txtmapsthefeatures

totheirindexes.Forexa