helpful equations to know

SSE

\(SSE=\sum_{i=1}^{n}|\hat{Y}_{i}-b_0|\)

\(SSE_A=(\sum_{i=1}^{n}|Y_{i}-b_0|)^2*n-1\) only when it is a simple model aka when MSE=variance and RMSE is standard deviation

so: \(SSE_A=SD*n-1\)

SSR

with SSE

\(SSR= SSE_C - SSE_A\)

with data points aka STARBURST

\(SSR=\sum_{i=1}^{n}(\hat{Y}_{iC}-\hat{Y}_{iA})^2\)

\(SSR=n(\hat{Y}_{iC}-\hat{Y}_{iA})^2\)

MSE

with SSE

\(MSE= \frac{\sum|\hat{Y}_{i}-b_0|} {n-P_A}\)

or

\(MSE= \frac{SSE} {df}\)

in the simple model MSE= variance!!!

RMSE

\(RMSE=\sqrt(MSE)\)

**in a simple model, MSE= variance (\(S^2\)), RMSE= standard deviation (s)

PRE

with SSE

\(PRE=\frac{ SSE_C - SSE_A} {SSE_C}\)

with SSR

\(PRE=\frac{SSR} {SSE_C}\)

F stat

with PRE

\(F_{(P_a-P_c, n-P_a)}=\frac{\frac{PRE}{P_a-P_c}}{\frac{1-PRE}{n-P_a}}\)

with SSE

\(F_{(P_a-P_c, n-P_a)}=\frac{\frac{SSR}{P_a-P_c}}{\frac{SSE_a}{n-P_a}}\)

with MS

\(F_{(P_a-P_c, n-P_a)}=\frac{MSR} {MSE}\)

T stat

with mean and null

\(t_{(n-1)}=\frac{\overline{Y}-B_o} {\frac{S_Y}{\sqrt(n)}}\)

with F

\(t_{(n-1)}=\sqrt(F*)\)

confidence interval

with F and MSE

\({b_{0}}\:\pm \sqrt{(F_{crit;1,n-P_A;\alpha} \: * \: (MSE))/n}\)

with PRE and \(SSE_C\)

\({b_{0}}\:\pm \sqrt{(PRE_{crit;1,n-P_A;\alpha} \: * \: (SSE_C))/n}\)

with T crit

\({b_{0}}\:\pm T_{crit;1,n-P_A;\alpha} \: * \: \frac{RMSE}{\sqrt{n}}\)

\({b_{0}}\:\pm T_{crit;1,n-P_A;\alpha} \: * \: \frac{SD}{\sqrt{n}}\)

true PRE

\(1- ((1-PRE)*\frac{n-P_C} {n-P_A})\)

f2 (for pwr calc)

\(f2= \eta^2/(1-\eta^2)\)