Processing math: 100%

ANOVA

ANOVA

ANOVA

使用時機

treatment 1treatment 2treatment kx11x12x1kx21x22x2kxn11xn22xnkk¯x1¯x2¯xkˉˉx

公式

SSE=kj=1nji=1(xij¯xj)2=kj=1(nj1)s2jSSTR=kj=1nj(¯xjˉˉx)2SST=kj=1nji=1(xijˉˉx)2

圖形

SSTSSESSTR

表格

source of variationsum of squaresdegrees of freedommean squareF
TreatmentsSSTRk1MSTR=SSTRk1F=MSTRMSE
ErrorSSEnkMSE=SSEnk
TotalSSTn1

與Simple Linear Regression E(y)=β0+β1x比較一下

source of variationsum of squaresdegrees of freedommean squareF
RegressionSSR1MSR=SSR1F=MSRMSE
ErrorSSEn2MSE=SSEn2
TotalSSTn1

Multiple Regression E(y)=β0+β1x1+β2x2++βpxp.

source of variationsum of squaresdegrees of freedommean squareF
RegressionSSRpMSR=SSRpF=MSRMSE
ErrorSSEnp1MSE=SSEnp1
TotalSSTn1

No comments:

Post a Comment