{[ promptMessage ]}

Bookmark it

{[ promptMessage ]}

IOE366-Ch13-Multiple+Regression

# IOE366-Ch13-Multiple+Regression - Ch13 Nonlinear and...

This preview shows pages 1–6. Sign up to view the full content.

Ch13: Nonlinear and Multiple Regression 13.1 Aptness of Model (Checking Assumptions) 13.2 Regression with Transformed Variables. 3.3 Polynomial Regression 13.3 Polynomial Regression 13.4 Multiple Regression 3 5 Issues in Multiple Regression 13.5 Issues in Multiple Regression (… the Kitchen Sink) 13.1 Aptness of the SLR Model L i n e a r i t y 1. Linearity 2. Homogeneity 3. Normality 4. Independence ε i ~ NID (0, σ 2 ) 5. Model Omissions 6. Outliers 2 Standardized Residuals i ˆ Recall e , is called the ith residual. i s(rv) there spon sea tx ii yy =− i Y is ( r.v.), the response at x , ˆ is (r.v.), the estimated response at x i Y then, e i is a random variable with mean 0 and variance aac e 2 2 ˆ () 1 ˆ 1 xx V YY σ ⎡⎤ −= 3 nS ⎢⎥ ⎣⎦ Standardized Residuals ± The standardized residuals are * 2 ˆ 1,2,. .., ˆ 1 y ei n == 2 1 −− 4

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
MINITAB OUTPUT (regular residuals) Residuals Versus the Fitted Values (response is Won) 4 3 2 1 0 -1 Residual 2 -2 -3 5 12 7 2 Fitted Value MINITAB OUTPUT (standardized residuals) Residuals Versus the Fitted Values (response is Won) 2 1 sidual 0 dardized Re -1 -2 Stan 6 12 7 2 Fitted Value Graph: versus ˆ i y 14 0 Scatterplot of Y (Games Won) vs Y hat 12 10 n) 8 6 Y (Games Wo 4 2 0 0 7 12 10 8 6 4 2 0 Y hat Residuals versus ˆ Residuals Versus the Fitted Values (response is Games Wo) 2 1 esidual 0 ndardized R -1 -2 Sta 8 12 7 2 Fitted Value
Residuals versus Observation # Residuals Versus the Order of the Data (response is Games Wo) 2 1 esidual 0 andardized R 25 20 15 10 5 -1 -2 St 9 Observation Order … Four Pack 99 5.0 Normal Probability Plot Versus Fits Residual Plots for Y (Games Won) 90 50 10 Percent 2.5 0.0 -2.5 Residual 5.0 2.5 0.0 -2.5 -5.0 1 Residual 12 9 6 3 -5.0 Fitted Value Histogram Versus Order 8 6 4 2 Frequency 5.0 2.5 0.0 -2.5 10 4 2 0 -2 -4 0 Residual 28 26 24 22 20 18 16 14 12 10 8 6 4 2 -5.0 Observation Order 1. Linearity (…Would a better function fit the data?) To check for “linearity”, we simply need more than one observation at one or more levels of x. Compute 2 j s and y for each level of x with repeated bserva ons observations. 11 We have two estimates of σ 2 : () 2 11 n c PE ij ji SS == =− ∑∑ ( ) 1 1 df N = = / MS = 12 2 EMS σ =

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
From the ANOVA Table we obtained: () 2 1 ˆ j n c Ei ij PE LF SS y =− = + ∑∑ 11 ji == 2 E df N / EE MS = 13 The pure error estimate of σ 2 , , should be smaller than since: SS E = SS PE + SS LF Sum of Squares ck of it Lack of Fit 14 2 ˆ 1 = whats left = (N-2) -(N-c) c-2 / = ( ) 2 01 1 2 jj nEy x EMS ββ σ = ⎡⎤ −+ ⎣⎦ =+ 15 2 ( ) ( ) ( ) L F If E MS Ey = If another model is more apt ()( ) LF →≥ 16
To test the linearity hypothesis ( ) ( ) 00 1 10 1 H:±± H: jj Ey x ββ =+ ≠+ 2 , ) ~ LF MS F −− (2 cNc PE Reject H 0 if . .. MINITAB automatically conducts this test when plicates are available for some levels of X 17 replicates are available for some levels of X. Example: Strength vs Additive Fitted Line Plot Strength = 10.94 + 0.1640 % Additive 25 20 S 5.12174 R-Sq 5.3% R-Sq(adj) 1.2% 15 Strength 10 5 18 35 30 25 20 15 % Additive Example: F-Test for LOF Results for: Additive.MTW Regression Analysis: Strength versus % Additive he regression equation is The regression equation is Strength = 10.9 + 0.164 % Additive Predictor Coef

This preview has intentionally blurred sections. Sign up to view the full version.

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

{[ snackBarMessage ]}

### Page1 / 32

IOE366-Ch13-Multiple+Regression - Ch13 Nonlinear and...

This preview shows document pages 1 - 6. Sign up to view the full document.

View Full Document
Ask a homework question - tutors are online