This preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: Point Estimation population 2 , σ μ sample 2 , S X unknown parameters observed sample statistics estimation X = μ 2 2 S = σ estimators Point Estimation Parameter : θ Estimator : θ Exampl e ( 29 θ , ~ ,..., , 2 1 U X X X iid n Random sample from uniform ( 29 n X X X ,..., , max ˆ 2 1 = θ Reasonable to use maximum of sample to estimate the upper bound Also reasonable to use sample mean to estimate population mean X 2 ˆ = θ Which estimator is better? Bias ( 29 ( 29 θ θ θ = E bias ( 29 θ bias overestimate ( 29 < θ bias underestimate ( 29 = θ bias unbiased Example : Random sample ( 29 μ μ all for = X E μ of estimator unbiased is X ( 29 2 2 2 all for σ σ = S E 2 2 of estimator unbiased is σ S Bias Exampl e ( 29 θ , ~ ,..., , 2 1 U X X X iid n Random sample from uniform X 2 ˆ 1 = θ ( 29 n X X X ,..., , max ˆ 2 1 2 = θ ( 29 θ θ = 1 ˆ E Unbiased ( 29 θ θ 1 ˆ 2 + = n n E Biased ( 29 θ θ θ θ 1 1 1 ˆ 2 + = + = n n n bias underestimate ( 29 n X X X n n ,..., , max 1 ˆ 2 1 3 + = θ is unbiased Mean Square Error unbiased small variance unbiased large variance biased small variance biased large variance Mean Square Error θ target 1 θ unbiased is 1 θ 2 θ biased is 2 θ 1 2 n better tha is But θ θ ( 29 ( 29  = 2 ˆ θ θ θ E MSE ( 29 ( 29 2 θ θ bias Var + = ( 29 ( 29 1 2 θ θ MSE MSE < Efficiency 2 1 to compared of efficiency θ θ ( 29 ( 29 ( 29 1 2 2 1 , θ θ θ θ MSE MSE eff = ( 29 1 , if than efficient more is 2 1 2 1 θ θ θ θ eff ( 29 ( 29 ( 29 unbiased. are ˆ and ˆ both if , 2 1 1 2 2 1 θ θ θ θ θ θ Var Var eff = Efficiency Exampl e ( 29 θ , ~ ,..., , 2 1 U X X X iid n Random sample from uniform X 2 ˆ 1 = θ ( 29 n X X X ,..., , max ˆ 2 1 2 = θ Unbiased Biased ( 29 θ θ 1 1 ˆ 2 + = n bias ( 29 ( 29 ( 29 n n X Var Var MSE 3 12 4 2 ˆ ˆ 2 2 1 1 θ θ θ θ = × = = = ( 29 ( 29 ( 29 2 1 ˆ 2 2 2 + + = n n n Var θ θ ( 29 ( 29 ( 29 ( 29 ( 29 2 1 2 2 1 1 ˆ 2 2 2 2 2 + + = + + + + = n n n n n n MSE θ θ θ θ ( 29 ( 29 ( 29 ( 29 ( 29 2 for 1 6 2 1 ˆ ˆ ˆ , ˆ 2 1 1 2 + + = = n n n MSE MSE eff θ θ θ θ efficient more is 2 θ Efficiency Exampl e ( 29 θ , ~ ,..., , 2 1 U X X X iid n Random sample from uniform ( 29 n X X X ,..., , max ˆ 2 1 2 = θ ( 29 ( 29 ( 29 2 1 ˆ 2 2 2 + + = n n n Var θ θ ( 29 ( 29 ( 29 2 1 2 ˆ 2 2 + + = n n MSE θ θ ( 29 ( 29 ( 29 1 for 1 1 2 ˆ ˆ ˆ , ˆ 3 2 2 3 + = = n n n MSE MSE eff θ θ θ θ efficient more is 3 θ ( 29 n X X X n n ,..., , max 1 ˆ 2 1 3 + = θ Unbiased ( 29 ( 29 ( 29 ( 29 2 ˆ 1 ˆ ˆ 2 2 2 3 3 + = + = = n n Var n n Var MSE θ θ θ θ Consistency ( 29 1 ˆ lim 0, any for if of estimator consistent a is = < ∞ → ε θ θ ε θ θ n n n P n θ ˆ Estimator of θ based on a sample of size...
View
Full
Document
 Fall '10
 MrChung
 Statistics, Normal Distribution, Maximum likelihood, Estimation theory, Likelihood function

Click to edit the document details