0
$\begingroup$

I am wondering what exactly means the intercept in the following? Is it a mean? I was told that is the mean of first category. Is it?

Call:
glm(formula = Pollution ~ Temp + Industry + Population + Wind + Rain + Wet.days, data = Pollute)

Deviance Residuals: 
    Min       1Q   Median       3Q      Max  
-23.159   -8.522   -1.147    5.823   48.586  

Coefficients:
             Estimate Std. Error t value Pr(>|t|)    
(Intercept) 111.59357   47.11131   2.369 0.023673 *  
Temp         -1.26510    0.61796  -2.047 0.048427 *  
Industry      0.06542    0.01567   4.175 0.000196 ***
Population   -0.03975    0.01506  -2.640 0.012430 *  
Wind         -3.17961    1.81397  -1.753 0.088643 .  
Rain          0.50507    0.36118   1.398 0.171054    
Wet.days     -0.04911    0.16123  -0.305 0.762554    
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

(Dispersion parameter for gaussian family taken to be 212.3861)

    Null deviance: 22037.9  on 40  degrees of freedom
Residual deviance:  7221.1  on 34  degrees of freedom
AIC: 344.37

Number of Fisher Scoring iterations: 2
Improve this question
$\endgroup$
3
  • $\begingroup$ Since you're just fitting a Gaussian model, you would call this model a "multiple regression". Are you interested in the interpretation for GLMs more generally, or specifically for the Gaussian multiple regression model? $\endgroup$
    – Glen_b
    Commented Feb 7, 2015 at 3:04
  • $\begingroup$ Thanks Glen_b. It the definition and application of intercept different in multiple regression and GLMs? I know that the intercept is when our x in the regression is zero. I just don't understand for instance in my model from what data points it is produced and what is telling me? $\endgroup$
    – Farshad
    Commented Feb 7, 2015 at 5:04
  • 2
    $\begingroup$ Well, the interpretation is certainly simpler in multiple regression, given that nice identity link. The interpretation in the case of a more general GLM involves and extra step of explanation. If you only need it for the Gaussian case, you should ask about that case rather than asking for the more complicated explanation required to cover GLMs in general $\endgroup$
    – Glen_b
    Commented Feb 7, 2015 at 5:36

0

Reset to default