Está en la página 1de 1

1/20/2017 archive.ics.uci.edu/ml/machinelearningdatabases/lungcancer/lungcancer.

names

1.Title:LungCancerData

2.SourceInformation:
Datawaspublishedin:
Hong,Z.Q.andYang,J.Y."OptimalDiscriminantPlaneforaSmall
NumberofSamplesandDesignMethodofClassifieronthePlane",
PatternRecognition,Vol.24,No.4,pp.317324,1991.
Donor:StefanAeberhard,stefan@coral.cs.jcu.edu.au
Date:May,1992

3.PastUsage:
Hong,Z.Q.andYang,J.Y."OptimalDiscriminantPlaneforaSmall
NumberofSamplesandDesignMethodofClassifieronthePlane",
PatternRecognition,Vol.24,No.4,pp.317324,1991.
Aeberhard,S.,Coomans,D,DeVel,O."Comparisonsof
ClassificationMethodsinHighDimensionalSettings",
submittedtoTechnometrics.
Aeberhard,S.,Coomans,D,DeVel,O."TheDangersof
BiasinHighDimensionalSettings",submittedto
patternRecognition.

4.RelevantInformation:
ThisdatawasusedbyHongandYoungtoillustratethe
poweroftheoptimaldiscriminantplaneeveninillposed
settings.ApplyingtheKNNmethodintheresultingplane
gave77%accuracy.However,theseresultsarestrongly
biased(SeeAeberhard'ssecondref.above,oremailto
stefan@coral.cs.jcu.edu.au).Resultsobtainedby
Aeberhardetal.are:
RDA:62.5%,KNN53.1%,Opt.Disc.Plane59.4%

Thedatadescribed3typesofpathologicallungcancers.
TheAuthorsgivenoinformationontheindividual
variablesnoronwherethedatawasoriginallyused.

Intheoriginaldata4valuesforthefifthattributewere1.
Thesevalueshavebeenchangedto?(unknown).(*)
Intheoriginaldata1valueforthe39attributewas4.This
valuehasbeenchangedto?(unknown).(*)


5.NumberofInstances:32

6.NumberofAttributes:57(1classattribute,56predictive)

7.AttributeInformation:

attribute1istheclasslabel.

Allpredictiveattributesarenominal,takingoninteger
values03

8.MissingAttributeValues:Attributes5and39(*)

9.ClassDistribution:
3classes,
1.) 9observations
2.) 13"
3.) 10"

http://archive.ics.uci.edu/ml/machinelearningdatabases/lungcancer/lungcancer.names 1/1