王静龙《非参数统计分析》课后计算题参考的答案解析

王静龙《非参数统计分析》课后习题计算题参考答案 习题一1. One Sample t-test for a MeanSample Statistics for xN Mean Std. Dev. Std. Error26 1.38 8.20 1.61Hypothesis TestNull hypothesis: Mean of x = 0Alternative: Mean of x A= 0t Statistic Df Prob > t0.861 25 0.397695 % Confidence Interval for the MeanLower Limit: -1.93Upper Limit: 4.70则接受原假设 认为一样习题二1.描述性统计乘机服务机上服务机场服务三者平均平均79.78平均54.46平均58.48平均64.24标准误差1.174661045标准误差2.085559681标准误差2.262605标准误差1.016376中位数82中位数55.5中位数58.5中位数64.5众数72众数60众数52众数65标准差8.306107908标准差14.74713393标准差15.99903标准差7.186861方差68.99142857方差217.4779592方差255.969方差51.65098峰度-1.059134152峰度0.083146927峰度0.41167峰度-0.04371偏度-0.164016852偏度0.264117712偏度-0.26232偏度-0.08186区域32区域65区域76区域34.66667最小值63最小值25最小值16最小值44最大值95最大值90最大值92最大值78.66667求和3989求和2723求和2924求和3212观测数50观测数50观测数50观测数50置信度(95.0%)2.360569749置信度(95.0%)4.191089091置信度(95.0%)4.546874置信度(95.0%)2.042483习题三1.1S+=13 n = 39H : me = 6500 H : me < 650001P {S +< 13}二BINOMDIST(13,39,0.5,1)=0.026625957另外:在 excel2010 中有公式 BINOM.INV(n,p,a) 返回一个数值,它使得累计二项式分布的函数值大于或等于临界值 a 的最小整数m = inf
)丿JS+ = 13 < d = 13以上两种都拒绝原假设,即中位数低于65001.2c = inf 2 c*m = inf 2 m*I ii=0BINOM.INV(40,0.5,1-0.025)=26 d=n-c=40-26=14 x = 5800 x = 640014 26me = x = 6200202.S+=40 n =70H : me = 6500 H : me 主 65002 P(S + > 40)= 2*(1-BINOMDIST(39,70,0.5,1))=0.281978922则接受原假设,即房价中位数是65003.1S+=1552n = 1552 + 527 = 2079H 0: P = 1n比较大,P(S + > 1552人1H : p > —1 2则用正态分布近似f 1039.5-1552+0.5、,7519.75 丿=5.33E-112另外:S+=1552 n = 1552 + 527 = 2079、a2c = inf 2 c * :m = inf 2 m* :11Y 工 In ]i2丿.Ii丿i =c*f 1丿n迟In丿l2 丿 i=01 i 丿m=BINOM.INV(2079,0.5,0.975)=1084则拒绝原假设,即相信孩子会过得更好的人多3.2P 为认为生活更好的成年人的比例,则1522p的比估计是::—2 =0.74651320794.S += 18154 n = 157860H : P = 65 H : P > 650 0.906 1 0.906p = 1 - 0.906 = 0.094S+ ~ b(n, p)P(S + > 18154)= 1 - BIN0MDIST(18153,157860,0.094,1)=0因为 0〈0.05 则拒绝原假设习题四1.车辆添加剂1添加剂2差值符号差的绝对值绝对值的秩122.3221.251.07+1.076225.7623.971.79+1.798324.2324.77-0.54-0.543421.3519.262.09+2.0910523.4323.120.31+0.311626.97260.97+0.974718.3619.4-1.04-1.045820.7517.183.57+3.5712924.0722.231.84+1.8491026.4323.353.08+3.08111125.4124.980.43+0.4321227.2225.91.32+1.327符号秩和检验统计量:W+ =6+8+10+1+4+12+9+11+2+7=70=650.025p 值为 2P(W+ > 70 ),当 n=12得 c 所以p值小于2P(W+ > 65 )=0.05即拒绝原假设被调查者x符号绝对值一个随机秩平均秩110+0004-1112.561+122.5211+132.524-1142.512+2575-226782+277152+287252+297143+31010.519-331110.524+4121474+41314124+41414174+41514224+4161495+51717.5105+51817.5188+819192611+1120203-1313212113-141422221615+1523232016+16242423-23232525符号秩和检验统计量:W+ =2.5+2.5+7+7+7+7+10.5+14+14+14+14+14+17.5+17.5+19+20+23+24=234.5=2360.025p值为2P(W+ > 234.5),当n=25 得c 所以p值小于2P(W+ > 236)=0.05即接受原假设符号检验:S + = 18 n = 26H : me = 0 H : me 丰 02 P(S + > 18)= 2*(1-BINOMDIST(17,25,0.5,1))=0.043285251则拒绝原假设t检验:t 统计量=0.861 df=25 p=0.3976接受原假设3.零售店豪华车普通车差值差值-100绝对值秩139027012020205239028011010102345035010000438030080-20205540030010000639034050-50508735029060-40407840032080-20205937028090-1010210430320110101021)W+=5+2+2=9 n = 8查表可得:c = 330.025, n(n +1) d = - c0.025 2 0.0252 P(W + < 3) = 0.052P(W + < 9) > 0.05Walsh 平均由小到大排列:50 55 60 65 65 70 7085 90 90 90 90 90 9070 75 75 7595 95 95 95则 接受原假设2)零售店豪华 车普通 车差值Walsh 值 Walsh= (x + x )/2 1 < i < j < ni ji=1i=2i=3i=4i=5i=6i=7i=8i=9i=10139027012012023902801101151103450350100110105100438030080100959080540030010011010510090100639034050858075657550735029060908580708055608400320801009590809065708093702809010510095859570758590104303201101151101059510580859510011080 80 80 80 80 80 80 85 85 85 8595 95 100 100 100 100 100 100 100 105 105105 105 105 110 110 110 110 110 115 115 120AN=55则对称中心为0二W 二W = 90((N+l)/2) 28d 二n(n +1)/4-0.5-U "(n + l)(2n +1)/24 二27.5-0.5-1.96x、,;10xllx21/24 二7.77101146c 二 n(n +1)/4 + 0.5 + U ':n(n + 1)(2n +1)/24 二 27.5 + 0.5 +1.9610x11x21/24 二47.228988531-a/2 7因为c不是整数,则6 L介于W与w 之间,其中k表示比d大的最小整数即为8L ( k) ( k +1)A0 L为70与75之间,即为72.5则H-L的点估计为9095%的区间估计为【72.5,105]习题五1.1228002520026550265502690027350285002895029900301503045030450306503080031000313003135031350318003205032250323503275032900332503355033700339503410034800350503520035500356003570035900361003630036700372503740037750380503820038200388003920039700404004100050个和在一起的中位数是(33250+33550) /2=33400工资<33400元工资>33400元合计男职工N11=7N12=17N1+=24女职工N21=18N22=8N2+=26合计N+1=25N+2=25N=50p = 2 7 P(i, 24,25,50)二 0.005060988i=1p值很小,则拒绝原假设 即认为女职工的收入比男职工的低。
1.2Wilcoxon 秩和 W =1+2+3.5+5+6+7+8+10+11.5+11.5+13+15+16+女17.5+17.5+20+22+24+26+29+31+32+34+35+36+44.5=478 因为N=n+m=50,查不到表,则用其渐进正态分布求解 n=26,m=24,N=50,W = 478女( )(W -n(N +1)/2)p=2^W+ < 478丿=® 产 一 =0.000327643(Jmn(N +1)/12 丿则拒绝原假设,认为女职工的收入低2.Wilcoxon 秩和 W =1+2+3+4+5+6+8+10+12=51B因为N= n+m=19,则用其渐进正态分布求解n=9,m=10,N=19, W = 51Bp=2P(W+ < 51)=2*©f n( N +1)/2 [=0.001450862(Jmn(N +1)/12 丿则拒绝原假设,认为A比B的作用好7.指数1116(11 月)1120(12 月)11251125113011471149114911511152秩12345678910指数1155116111661169117111761182118411841194秩11121314151617181920Wilcoxon 秩和 W =1+5+7+8+10+11+16+18+19+20=11511月因为 n二n +n =20,12n =10,n =10,n=2O, W = 115 1 1 11 月所以,p=2P(W+ > 115丿查表可得:c =1580.025因为,158>115,所以p值一定大于0.05则接受原假设,认为11月和12月的波动相同位置参数检验:d = nm/2 - 0.5 - U v'nm( N +1)/12 = 10*10/2 - 0.5 -1.96,10*10*101/12 = 23.641-a/2 c = nm/2 + 0.5 + U vnm( N +1)/12 = 10*10/2 + 0.5 +1.9^.10*10*101/12 = 76.361-a/20L介于w与w 之间,其中k表示比d大的最小整数即为24 0L =-29L (k) (k +1) L0n介于w 与⑷之间,其中[c]表示比c小的最大整数即为76 0U =17U [c]-1 [c] U所以,区间为:[-29,17] 即 0 在区间内则认为11月和12 月的波动相同8机器(y)平均秩myay5.053344135.1665.5342.255.55.198825685.21210.5182.2510.55.221514100145.251716.556.2516.5尺度参数检验:M =艺 a(R ) = 2269.25yii=1E(M )= nyN 2 -112工12 (u 3 -T )— i=1——i 」12 N=22*472 -1 288 ~12__ 12*47=4036.765957D(M )= (nm )「工12 t (d -(N +1)/2》-N(N2 —1)/144 y N (N — 1儿 t=1 t t=22*25【796056.5-47*(47*47-1)/144]= 202328.7273 47*46所以渐进正态分布计算其p值:'M -E (M )'y / y| Jp=0则 认为较小nm=4.25641E-05A =艺 a(R ) = 346.5yii=1E(A )=(N+1)24N工12a 3 -t)]■i=1 1 i12 N=22* 勢 一 1147 =2583829787D(M )=nmN(N -1)工12 t d 2 -t =1 t t(N+1)416N=22*25 24245--^^ =6167.005039 47*46 L 16*47所以渐进正态分布计算其p值:[A -E(A ”p=®| T y I =0.869085147〔丙丿则认为较大 综合:因为M比较小而A比较大,yy可知b丰1且b应该有bv1因为b