资源描述：
《data exploration and preparation》由会员上传分享，免费在线阅读，更多相关内容在工程资料-天天文库。
1、3DataexplorationandpreparationThischapterstartswithabriefsurveyofthedifferenttypesofdata.Itthengoesontodetailthetasksofexploring,analysingandpreparingthedatacollected,usingunivariate,bivariateandmultivariatemethodsandstatisticaltests.Thesetasksarefundamental,asthequalityoftheresult
2、sgenerallydependsmoreonthecaretakenherethanonthemodellingmethodwhichusestheseinputdata.Thechapterconcludeswithadiscussionofsampling.3.1ThedifferenttypesofdataThedatamaybequantitative,qualitative,or,forsomespecialproblems,textual.Quantitative(ornumerical)datamaybecontinuousordiscret
3、e.Continuous(orscaleorinterval)dataarethosewhosevaluesbelongtoaninfinitesubsetofthesetRofrealnumbers(e.g.wages,amountofpurchases).DiscretedataarethosewhosevaluesbelongtoafiniteorinfinitesubsetofthesetNofnaturalintegers(e.g.numberofchildren,numberofproductsbought).Whenthesetofvaluesisi
4、nfinite,orjustlarge(severaltens),discretedataareconsideredtobecontinuousdata(e.g.age).Qualitative(orcategorical)dataaredatawhosesetofvaluesisfinite.Thesevaluesarenumericaloralphanumeric,butwhentheyarenumericaltheyaremerelycodes,notquantities(e.g.socio-occupationalcategory,departmentn
5、umber).Thevaluesoftextdataareuncodedtexts,writteninnaturallanguage:examplesarelettersofcomplaint,reportsandpressdespatches.Whatdistinguishescontinuousanddiscretedatafromothertypesisthattheyareconcernedwithquantities,sowecanperformarithmeticaloperationsonthem;moreover,theyareordered
6、(wecancomparethembyanorderrelationship‘’).Qualitativedataarenotquantities,buttheymaybeordered;inthiscasewespeakofordinalqualitativedata(e.g.‘low,medium,high’).Non-orderedqualitativedataarecallednominal.Ordinaldatacanbeclassedinthefamilyofdiscretedataandtreatedinthesameway.Theanaly
7、sisoftextdatarequiresamixtureoflinguisticsandstatistics:thisistextmining,whichisdiscussedinChapter14.NotethattextdatainnaturallanguagemaycontainellipticalDataMiningandStatisticsforDecisionMaking,FirstEdition.Ste´phaneTuffe´ry.Ó2011JohnWiley&Sons,Ltd.Published2011byJohnWiley&Sons,Lt
8、d.44DATAEXPLORATIONANDPREP
