# the positive constant is called the 1 100 condent

This preview shows page 1. Sign up to view the full content.

This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: to be ﬁnite. When should we stop increasing n? Till there is enough guarantee that Yn must be suﬃciently close to p ! M. Chen (IE@CUHK) ENGG2430C lecture 10 20 / 29 ATV poll problem revisit: Conﬁdence interval * Question: Given positive constants 0 &lt; δ &lt; 1 and ε , how large should n be, to make sure with at most probability δ , Yn ∈ (p − ε , p + ε )? / That is, ﬁnd the smallest n so that P (Yn ∈ (p − ε , p + ε )) = P (|Yn − p | ≥ ε ) ≤ δ . / The positive constant ε is called the (1 − δ ) × 100%-conﬁdent interval. Exercises: Prove that for any p ∈ [0, 1], p (1 − p ) ≤ 1 . 4 Compute a suitable n for δ = 0.01 and ε = 0.05. (Hint: use Chebyshev inequality.) M. Chen (IE@CUHK) ENGG2430C lecture 10 21 / 29 ATV poll problem revisit: Conﬁdence interval 1 Solution: By Chebyshev inequality and p (1 − p ) ≤ 4 , we obtain an upper bound for P (|Yn − p | ≥ ε ) as P (|Yn − p | ≥ ε ) ≤ = Var(Yn ) ε2 p (1 − p ) 1 ≤ 2 nε 4nε 2 A suﬃcient condition for P (|Yn − p | ≥ ε ) ≤ δ i...
View Full Document

Ask a homework question - tutors are online