Y i = β 1 + β 2 X i + U i E ( U | X ) = 0 V a r ( U | X ) = V a r ( Y | X ) = σ 2 Y i = β 1 + β 2 X i + U i Y i ^ = β 1 ^ + β 2 ^ X i U i ^ = Y i − Y i ^ {\displaystyle {\begin{aligned}&Y_{i}=\beta _{1}+\beta _{2}X_{i}+U_{i}\\&E(U|X)=0\\&Var(U|X)=Var(Y|X)=\sigma ^{2}\\&Y_{i}=\beta _{1}+\beta _{2}X_{i}+U_{i}\\&{\widehat {Y_{i}}}={\widehat {\beta _{1}}}+{\widehat {\beta _{2}}}X_{i}\\&{\widehat {U_{i}}}=Y_{i}-{\widehat {Y_{i}}}\\\end{aligned}}}
使用 β 1 {\displaystyle \beta _{1}} 或 β 0 {\displaystyle \beta _{0}} 来表示 Y 轴截距完全是自行决定的。
β 2 ^ = ∑ ( Y i − Y ¯ ) ( X i − X ¯ ) ∑ ( X i − X ¯ ) 2 β 1 ^ = Y ¯ − β 2 ^ X ¯ V a r ( β 2 ^ ) = σ 2 ∑ ( X i − X ¯ ) 2 V a r ( β 1 ^ ) = ∑ X i 2 ∗ σ 2 n ∗ ∑ ( X i − X ¯ ) 2 σ 2 ^ = ∑ U i 2 ^ n − 2 {\displaystyle {\begin{aligned}&{\widehat {\beta _{2}}}={\frac {\sum {(Y_{i}-{\bar {Y}}})(X_{i}-{\bar {X}})}{\sum {(X_{i}-{\bar {X}})^{2}}}}\\&{\widehat {\beta _{1}}}={\bar {Y}}-{\widehat {\beta _{2}}}{\bar {X}}\\&Var({\widehat {\beta _{2}}})={\frac {\sigma ^{2}}{\sum {(X_{i}-{\bar {X}})^{2}}}}\\&Var({\widehat {\beta _{1}}})={\frac {\sum {X_{i}^{2}}*\sigma ^{2}}{n*\sum {(X_{i}-{\bar {X}})^{2}}}}\\&{\widehat {\sigma ^{2}}}={\frac {\sum {\widehat {U_{i}^{2}}}}{n-2}}\\\end{aligned}}}
U 和 ϵ {\displaystyle \epsilon } 都被用来表示误差项。
S 2 2 = V a r ( β 2 ^ ) ^ = σ 2 ^ ∑ ( X i − X ¯ ) 2 S 1 2 = V a r ( β 1 ^ ) ^ = ∑ X i 2 ∗ σ 2 ^ n ∗ ∑ ( X i − X ¯ ) 2 S . E . ( β 2 ^ ) = V a r ( β 2 ^ ) ^ S . E . ( β 1 ^ ) = V a r ( β 1 ^ ) ^ {\displaystyle {\begin{aligned}&S_{2}^{2}={\widehat {Var({\widehat {\beta _{2}}})}}={\frac {\widehat {\sigma ^{2}}}{\sum {(X_{i}-{\bar {X}})^{2}}}}\\&S_{1}^{2}={\widehat {Var({\widehat {\beta _{1}}})}}={\frac {\sum {X_{i}^{2}}*{\widehat {\sigma ^{2}}}}{n*\sum {(X_{i}-{\bar {X}})^{2}}}}\\&S.E.({\widehat {\beta _{2}}})={\sqrt {\widehat {Var({\widehat {\beta _{2}}})}}}\\&S.E.({\widehat {\beta _{1}}})={\sqrt {\widehat {Var({\widehat {\beta _{1}}})}}}\\\end{aligned}}}
S 2 {\displaystyle S^{2}} 用于表示样本方差,S.E. 表示标准误。
T S S = ∑ ( Y i − Y ¯ ) 2 E S S = ∑ ( Y i ^ − Y ¯ ) 2 R S S = ∑ U i 2 ^ {\displaystyle {\begin{aligned}&TSS=\sum {(Y_{i}-{\bar {Y}})^{2}}\\&ESS=\sum {({\widehat {Y_{i}}}-{\bar {Y}})^{2}}\\&RSS=\sum {\widehat {U_{i}^{2}}}\\\end{aligned}}}
TSS 也可以表示为 SST,代表总平方和,ESS 可以表示为 SSE(误差),RSS 可以表示为 SSR(残差)。根据不同的文本,ESS 和 RSS 可能非常混乱,因为术语使用存在很大差异。
R 2 = E S S T S S = 1 − R S S T S S {\displaystyle R^{2}={\frac {ESS}{TSS}}=1-{\frac {RSS}{TSS}}}
log ( Y ) ^ = β 1 ^ + β 2 ^ log ( X ) {\displaystyle {\widehat {\log(Y)}}={\widehat {\beta _{1}}}+{\widehat {\beta _{2}}}\log(X)}