Question

a) Why is implementing a K-means clustering algorithm multiple times with a fixed K important to do? 119 b) Why is cross-validation preferred over resubstituting as a method to measure classification accuracy? Explain c) Give two situations when nearest neighbor classification may be preferred over linear and quadratic discriminant analysis methods in general. Explain your answer.

Answer 1

ustening algorithm nuJtipe times with a fived Kimpotant to do xity incneaser4 an od el ;nc neo.轟0.5 c.omelex;ㅓ Ci) Ted g eecupenalizes models that are too enous compe xiy3 e-fered over m e 凢e classifica 、ơn aceustaey i P novides a highvariones estimate of oat- -sa mfle ac cu ぐら K fold c几oss-val: dat,'on over t几a in // test. C.) spe it 사;从 Dut, is speed There are two 3ita a on s wh en e me model on the entise dataseA Test te model onthe a-tased elassificatn aecwnacy o coniect