class_10_15

# class_10_15 - Statistical Data Mining ORIE 474 Fall 2007...

This preview shows pages 1–5. Sign up to view the full content.

Statistical Data Mining ORIE 474 Fall 2007 Tatiyana Apanasovich 10/15/07 Generalized Linear Models & Logistic Regression

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

View Full Document
11.3 Generalized Linear Models (GLMs) Recall: Linear model (i) The Y(i) are independent random variables, with distribution N(μ(i), σ 2 ), where μ(i)=Σ j=0,. .,p a j x j (i) (ii) The parameters enter the model in a linear way via the sum ν(i)=Σ j=0,. .,p a j x j (i) (iii)The ν(i) and μ(i) are linked by ν(i)=μ(i) This permits 2 immediate generalizations in (i) and (iii), while retaining the advantages of (ii)
Generalizations of Linear Models to GLMs In (i): We can relax the requirement that Y(i) follow a normal distribution In (iii): We can generalize the link to a more general link function g : ν(i) = g(μ(i)) or Σ j=0,. .,p a j x j (i) = g(μ(i))

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

View Full Document
Ex: Logistic Regression Consider the following situation: The observed response in binary, that is y(i) in {0,1} Ex: Whether or not a person gets a certain question right Whether or not a person is diagnosed with a certain disease Assume: Probability that i th object yields 1 is p(i) Responses of different objects are independent This is the generalization of (i) above, that is, the Bernoulli distribution is replacing the normal distr.
This is the end of the preview. Sign up to access the rest of the document.

{[ snackBarMessage ]}

### Page1 / 14

class_10_15 - Statistical Data Mining ORIE 474 Fall 2007...

This preview shows document pages 1 - 5. Sign up to view the full document.

View Full Document
Ask a homework question - tutors are online