Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Chapter 2. Two-Variable Regression Analysis: Some Basic Ideas A Hypothetical Example Imagine a hypothetical country with a total population of 60 families. Question: To set a relationship between weekly family consumption expenditure (Y) and and weekly family income (X). Conditional Distribution of Y with respect to X Table gives the distribution of consumption expenditure Y corresponding What2.1 is Conditional Mean? How do you calculate it? Conditional Mean Y given 55(1/5)+60(1/5)+65(1/5)+70(1/5)+75(1/5) to a fixed level offor income X;that thatX=80: is conditional distribution of Y conditional = 65 upon the given values of X. Conditional Probabilities Conditional probability of Y given X. For P (Yexample: = 150Probability: / Xp =(Y=55 260) /=p(Y/X): X1/7 = 80) = 0,14 = 1/5 = 0,20 Conditional Mean or Conditional Expectation E (Y / X = Xi) It is read as “the expected value of Y given that X takes the specific value Xi” Data of Table 2.1 on a Plot Data of Table 2.2 on a Plot The Concept of Population Regression Function (PRF) If E (Y / X = Xi), then Therefore, E (Y / Xi) = f (Xi) PRF That is, if conditional mean of Y depends on each level of X variable, then conditional mean of Y is said to be a function of given X values. On PRF More PRF is E (Y / Xi) = f (Xi) Therefore, PRF is also linear function of Xi, that is: E (Y / Xi) = β1 + β2 Xi Where β1 and β2 are unknown parameters known as the regression coefficients. On PRF more Dependent Variable E (Y / Xi) = β1 + β2 Xi Intercept Slope Independent Variable Linear Population Regression Function Stochastic Specification of PRF E (Y / Xi) = β1 + β2 Xi ui = Yi – E (Y/Xi) Yi Expected Actual value of Y or estimated value =Stochastic E (Y/XError ) + Term u of Y with respect to X i i Then, Yi = β1 + β2 Xi + ui Stochastic Error Term E ( Yi / X) = E [E( Y / Xi)] + E (ui /Xi) E ( Yi / X) = E( Y / Xi) + E (ui /Xi) E (ui /Xi) = 0 Therefore, E ( Yi / X) = E( Y / Xi) The Sample Regression Function (SRF) PRF: Yi = β1 + β2 Xi + ui SRF: Yˆ = βˆ1 + βˆ 2 X i + uˆ i of β2 EstimatorEstimator of β1 Example on SRF PRF and SRF Compared PRF Error SRF Error