Processing math: 100%

Statistics Cheat Sheet

Statistics Cheat Sheet

Statistics Cheat Sheet

Simple Linear Regression

ˆy= b0+b1x
Coefficients:          
  Estimate Std. Error t value Pr(>|t|)  
(Intercept) (1)       ***
x (2) (3) (4)   ***

---
signif codes
Residual Standard error: (5)
Multiple R-squared: (6), Adjusted R-squared:
F-statistic: (7), p-value:
(1) b0
(2) b1
(3) ˆσb1
(4) t=b1/ˆσb1
(5) ˆσ
(6) r2
(7) F=MSRMSE
SSE, SSR, SST
Relation
Picture
SSE=(yiˆyi)2
SSR=(ˆyi¯y)2
SST=(yi¯y)2
SSE+SSR=SST
b0
b1
ˆσ
ˆσb1
r2
t
F
b0=¯yb1¯x
b1=(xi¯x)(yi¯y)(xi¯x)2
ˆσ=SSEn2
ˆσb1=ˆσ(xi¯x)2
r2=SSRSST
t=b1ˆσb1
F=MSRMSE
ANOVA Table E(y)=β0+β1x
source of variation sum of squares degrees of freedom mean square F
Regression SSR 1 MSR=SSR1 F=MSRMSE
Error SSE n2 MSE=SSEn2  
Total SST n1    
ANOVA Table E(y)=β0+β1x1+β2x2++βpxp
source of variation sum of squares degrees of freedom mean square F
Regression SSR p MSR=SSRp F=MSRMSE
Error SSE np1 MSE=SSEnp1  
Total SST n1    
Confidence Interval of ˆy at x=x0 b0+b1x0+tα/2,n2ˆσ1+1n+(x0¯x)2(xi¯x)2
Big Picture for SSE,SSR,SST,r2 and sx,sy,sxy,rxy  
SSE, SSTR, SST SSE=kj=1nji=1(xij¯xj)2=kj=1(nj1)s2j
SSTR=kj=1nj(¯xjˉˉx)2
SST=kj=1nji=1(xijˉˉx)2treatment 1treatment 2treatment kx11x12x1kx21x22x2kxn11xn22xnkk¯x1¯x2¯xkˉˉx
Picture SSTSSESSTR
ANOVA Table
source of variation sum of squares degrees of freedom mean square F
Treatments SSTR k1 MSTR=SSTRk1 F=MSTRMSE
Error SSE nk MSE=SSEnk  
Total SST n1    
Fisher's LSD
Test Statistic
Compare with
¯xi¯xjMSE(1ni+1nj)
tα/2,nk

No comments:

Post a Comment