This preview shows pages 1–5. Sign up to view the full content.
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 Document
Unformatted text preview: 1 Comparing Systems 13/10/2009 1 R. Jain book + M. Claypool [WPI] Sample versus Population1 • The word “sample” comes from the same root word as “example” word as example • Similarly, one sample does not prove a theory, but rather is an example • Basically, a definite statement cannot be made about characteristics of all systems 2 • Instead, make probabilistic statement about range of most systems – Confidence intervals 2 Sample versus Population2 • Say we generate 1billion random numbers ! mean µ and stddev σ – mean µ and stddev σ . – µ is population mean – That is, µ and σ are the “ True ” values • Put the numbers in a pool, draw sample of size n – Sample {x 1 , x 2 , …, x n } has mean , and stddev s – That is, and s are the “ measured ” values 3 • It is likely to have ≠ µ and s ≠ σ • Typically, µ is not known and may be impossible to know – Instead, get estimate of µ from Confidence Interval for the Mean • Obtain probability of µ in interval [c 1 ,c 2 ] Prob{c < µ < c } = 1 α – Prob{c 1 < µ < c 2 } = 1 α • (c 1 , c 2 ) is called confidence interval • α is called significance level • (1 α ).100% is called confidence level • Typically we want α small so that confidence level is high – Say, α =0.1 ⇒ 90% confidence level 4 ∴ the interval (c 1 , c 2 ) contains the value µ with 90% confidence level or with probability = 0.9 µ c 1 c 2 3 Determining Confidence Interval • 100(1 α ) % confidence interval for n observations is observations is: • t [ α /2; n1] = α /2entry in the ttable with n1 degrees of freedom (Student’s t distribution Table “Student” was an anonymous name 5 Table. Student was an anonymous name) 6 4 Example 1 • = 3.90, s = 0.95 and n = 30 • A 90% confidence interval for the mean ( α =0.1) = 3.90 ± 1.699 × 0.95 / = (3.61, 4.19) 30 3.9 3.9 4 1 1.9 2.7 2 8 (Sorted) CPU Time • We can state with 90% confidence that the population mean is between 3.61 and 4.19 The chance of error in this statement is 10%....
View
Full
Document
This note was uploaded on 10/27/2010 for the course CMPE 474 taught by Professor Muhammedsalamah during the Spring '10 term at Eastern Mediterranean University.
 Spring '10
 MUHAMMEDSALAMAH

Click to edit the document details