Questions tagged [link-function]
A transformation of a parameter governing a response distribution that is used as a crucial part of the generalized linear model to map that parameter's range (which may be from 0 to 1, or only positive values, e.g.) to the real number line $(-\infty, +\infty)$.
190
questions
0
votes
1
answer
26
views
How to prove the Poisson link function is a canonical link function?
So I'm a 3rd year undergraduate doing my thesisin football score models right now. In my thesis I want to include a proof of what the link function for the Poisson distribution is and why it relates ...
0
votes
0
answers
14
views
Difference between specifying link function in GLM and transforming y data in advance? [duplicate]
I'm a beginner in stats and I have a very basis question on GLM.
Suppose Gaussian distribution. GLM tries to do this:
$$g(E(y|X))=X\beta$$
where $g$ is a link function.
But if I transform $y$ before ...
- 101
0
votes
0
answers
57
views
Zero Inflated Beta Regression (ZOIB)
I'm using R to model zero inflated beta regression. Upon reading some documentation for the ZOIB package, I came across some functionality for adding random effects to one or many of the link ...
- 21
1
vote
0
answers
34
views
Is it possible to build residuals that are uncorrelated with predictors from regressions with a binary outcome, without an identity link?
Here: What do the residuals in a logistic regression mean? is a description of the different ways how one can generate residuals from a logistic regression.
My question concerns not the logistic ...
- 1,072
1
vote
1
answer
29
views
reasonable distributions for non-negative real-valued data with many zeroes, GAMs
Lots of data is real-valued, non-negative, and replete with 0s... for example any time count data are normalized by something: e.g., perhaps it makes sense normalize disease case counts taken over ...
- 349
1
vote
0
answers
18
views
Is it possible to average models with different link functions?
While I can compare models that have different link functions in terms of AIC/BIC weights, I think it's impossible to use those weights to create an averaged model. Am I right in believing this?
- 943
3
votes
1
answer
70
views
Under what conditions do different choices of link function for GLMs result in considerably different models?
The statistical folklore I have heard is that the choice of link function usually does not considerably affect the resulting fit of a GLM. For example, usually probit regression and logistic ...
- 1,213
1
vote
0
answers
85
views
Different results with different link functions
I have used the R package eventglm to construct pseudo observations, and want to estimate the relative risk and the risk difference for exposure with age adjustment....
- 11
2
votes
2
answers
94
views
Ascertaining GLM link visually
On pg. 125 in Agresti's Categorical Data Analysis, it's suggested by a plot of the dependent variable (a count) vs an independent variable (categorized version of continuous width variable) that the ...
- 325
3
votes
1
answer
192
views
Purpose for the conditions of Link Function
I am studying GLM at the moment and have a few questions regarding link functions.
Why are the conditions of the link function to be smooth monotonic function? What properties are preserved by having ...
2
votes
1
answer
21
views
Split linear predictors with link function
Is it possible to split linear predictors contribution up when talking glm of non-normal distributions?
If:
$$µ_i = g^{-1}(η_i)$$
and
$$µ_i = g^{-1}(β_0 + β_1X_{i1} + β_2X_{i2} +···+β_kX_{ik})$$
Is it ...
- 107
0
votes
1
answer
429
views
GLM with Gamma distribution: Choosing between two link functions
I need to perform a GLM based analysis on a purely positive, continuous, and highliy right skewed (inflated around low values) outcome variable. I tested several combinations of distributions and link ...
0
votes
1
answer
262
views
Back transform predict.gam() from nb link log model run?
I have model with 1 covariate. I would like to run y values from gam in another model. I used nb(log=link) in gam model. Because I used nb and link log in gam, do I need to back transform to use ...
0
votes
1
answer
145
views
Back transforming standard errors in a GLMM with a log-link
I'm using lme4 and have a GLMM with a log link and gaussian variance structure. I would like to report my fixed effect estimates with their standard errors, as well as the standard deviation of my ...
- 85
5
votes
1
answer
403
views
Which link function in binomial regression is better?
Concerning the choice of the link function in binomial regression (e.g. logit versus probit or cauchit), I wonder what the recommended comparison criterion might be. Note that I am not interested in ...
- 3,790
0
votes
0
answers
16
views
How can I use the link(g, lam) function from "psyphy" package to adjuste asymptotes?
I need to customize the asymptotes of the model, and I am trying with psyphy package which provides parameters for adjusting asymptotes in its ...
- 1
1
vote
1
answer
173
views
Transforming the expected value of $Y_i$ in binomial regression
Currently, I'm learning generalized linear regression (GLM). There is something troubling me concerning binomial regression.
In this text, in the part about the structure of a GLM, the random ...
- 113
4
votes
1
answer
125
views
Mean finger volume: Is a GLM with log link function appropriate?
I have a model where the volume ($V$) of a finger is normally distributed, with mean $\mu = \beta_0 L^{\beta_1}D^{\beta_2}$ (where $L=$ length, $D=$ diameter and $\beta_i \in \Bbb R$ for $i=0,1,2$) ...
- 41
0
votes
0
answers
228
views
Is this the correct way to compute confidence intervals on the original scale for GLM(M)s?
Suppose I have fitted a GLM and want to produce a confidence interval (or a prediction interval) on the original scale of the outcome. What I would do is estimate it on the link scale and then inverse ...
- 10.6k
3
votes
0
answers
79
views
Why is the canonical parameter linearly related to the input x in GLMs and why does it give the link function?
In Andrew Ng's CS229 notes, one of the three assumptions he makes for constructing GLM models is:
The natural parameter $\eta$ and the inputs x are related linearly: $\eta=\theta^Tx$
He goes on to ...
- 131
2
votes
1
answer
233
views
Link functions in poisson regression
I've recently started studying statistics and a question came up to my mind while reading about poisson regression:
If we have to exponentiate all terms in order to have only positive values, why do ...
- 305
5
votes
1
answer
2k
views
Interpretation difference between log link and log transformation
I have a question about the interpretation difference between log link of GLM and log transformation of LM.
I know that log transformation is for target variable but log link is for mean .But related ...
- 51
1
vote
0
answers
38
views
Is log-link function important in this case? [closed]
I have a positive count response $Y$ (number of times that a particular pattern was observed within a single day) and positive count independent variables $X_{1}, X_{2}$ (that are associated with the ...
- 2,040
0
votes
1
answer
128
views
Finding a confidence interval for difference of proportions
Let two independent random variables, $Y_1$ and $Y_2$ that have binomial distribution have parameters $n_1 = n_2 = 100$, $p_1$ and $p_2$, respectively, be observed to be equal to $y_1 = 50$ and $y_2 = ...
user298588
6
votes
1
answer
172
views
Help understand the virtue of generalized linear models
On page 4 of https://www.sagepub.com/sites/default/files/upm-binaries/21121_Chapter_15.pdf, the authors state the following strength of generalized models, which I don't quite understand.
Indeed, one ...
- 825
7
votes
1
answer
718
views
Which link function could be used for a glm where the response is per cent (0 - 100%)?
I am thinking about building a model (glm) where the response variable (y) is the cover (in per cent) of a plant species in a defined area, dependant of environmental variables.
However, I don't think ...
- 175
2
votes
1
answer
221
views
Conway-Maxwell-Poisson (CMP) - Coefficient interpretation (Log/IRR)
I'm using the Conway-Maxwell-Poisson (CMP) distribution to model the amount of nouns in a clause (data is under-dispersed). I've run the model using glmmTMB (family= "compois") but I'm ...
- 53
0
votes
0
answers
47
views
Why are we entitled to use the link function we prefer the most?
For a project, we have been trying to fit different models. When we used a Poisson regression, so a glm with a Poisson family, initially our fit was quite bad. But once we used the identity link ...
- 101
0
votes
1
answer
349
views
chi square GLM inference
Suppose at $m$ different positions on a line $a_1,....,a_m$, we sample from a i.i.d normal distribution $N(\mu_i,\sigma_i^2)$, $n_i$ times for each of the $1\le i\le m$ different points. Here of ...
- 33
2
votes
2
answers
497
views
Fit a zero-inflated Poisson GAM
I am trying to fit a zero-inflated Poisson GAM to my count data, and I want a log link. ziP() from the mgcv package does not ...
- 23
1
vote
1
answer
892
views
Appropriate link function for beta distribution
I am fitting a continuous proportion as my response in a beta distribution model.( I am using rstanarm to implement this model). For context the continuous proportion is the amount of time out of 30 ...
- 51
1
vote
0
answers
193
views
How does link function work in GLM? [closed]
I have several questions regarding the link function of generalized linear regression. I know how link function changes range of the distribution function's mean to the complete real line. But is that ...
- 11
3
votes
1
answer
172
views
Modeling both mean and variance in a linear model
I have a variable $X$ that decays log-normally with time, and I have estimated the mean and the SD of that log-linear relationship. I also have a (categorical) variable $Y$ which—I hypothesize—will ...
- 197
1
vote
2
answers
37
views
Link functions and interpreting credible intervals
I am pretty new to statistics, and was trying to interpret credible intervals from a bayesian analysis I had preformed. Some of my models are glms, and so have a link function. I know that to ...
- 51
3
votes
1
answer
1k
views
Log transformation in GLM and model fit
For a negative binomial GLM, are we allowed to write the log transformation in the following way?
...
4
votes
1
answer
4k
views
glm with log link in binomial family
I was reading this post
https://r-posts.com/simulations-comparing-interaction-for-adjusted-risk-ratios-versus-adjusted-odds-ratios/
and found that the author adjusted a glm with binomial family and ...
3
votes
1
answer
2k
views
Gamma GLM: why log-link is more common than canonical link
"The canonical link of Gamma GLM is $g(x)=1/x$ is often not very practical. Log-link is more appropriated in most cases." One reason I can think of is that log-link makes sure $\mu$, the ...
- 988
3
votes
1
answer
145
views
GLM: effect of link function on choice of transformation of covariate
It struck me that if I have data of the form below,
...
- 1,694
0
votes
1
answer
280
views
Should you standardize when using a Log link?
If I use a model with a log link function should I still standardize independent variables (since they differ in the scales range) or the log transformation is enough?
- 135
3
votes
2
answers
1k
views
Why do we choose exponential function as the nonlinearity in Possion GLM
In Poisson GLM, the response variable $Y$ follows the Poisson distribution
$$P(Y=y)=\lambda^y\exp(-\lambda)/y!$$
and:
$$\lambda=\exp(\bf \theta^Tx)$$
My question is why do we use exponential as the ...
- 211
8
votes
2
answers
5k
views
Generalized Linear Model and Identity link, what's its benefit?
I found a paper saying that a Generalized linear model with an identity link function was used. They standardize some continuous independent variable as well as the continuous dependent variable and ...
- 135
2
votes
1
answer
134
views
Generalised Linear Model help
Kind of new to coding using Rstudio here.
I have data for a survey for 600 individuals, over 4 years (150 p/year), for 30 categories being an absence or presence (0/1) resulting in a total score /30 ...
- 77
3
votes
0
answers
49
views
Is the following a GLM?
We know that the standard linear model is a partial case of the GLM scenario by taking the identity link function, i.e.
$$g(μ)=μ=η=x_i^Tβ$$
However, in one of our past papers we are asked to ...
- 374
2
votes
0
answers
69
views
Why does the canonical parameter give a link function? Why does this relate $E[Y]$ to $x^T \beta$?
If I have a pdf in the form $f(y|\theta,\phi)=\text{exp}\bigg(\frac{y\theta-b(\theta)}{a(\phi)}+c(y,\phi)\bigg)$, then $\theta$ is called the canonical parameter. I'm told we can get a link function $...
- 474
1
vote
1
answer
2k
views
Geometric distribution: finding canonical link and proving it is part of the natural exponential family?
Looking for some help on my statistics homework question!
The background to the question is: suppose that you toss a biased coin repeatedly (and independently) until you get a head. Let Y denote the ...
- 11
1
vote
1
answer
763
views
The identity link not used for binary response
Questions: The identity link is the standard one with normal responses but is not often used with binary or count responses. Why do you think this is?
My idea: The range for a linear predictor, and ...
13
votes
2
answers
747
views
Logistic regression with actual probabilities $\in(a,b)$ where $0<a<b<1$
When modelling probabilities with a logistic regression$^1$, the range of fitted probabilities is $(0,1)$. The logit function$^2$ asymptotes at $0$ and $1$, so this is a good match. However, in some ...
- 61.1k
4
votes
0
answers
67
views
Would this modification accelerate convergence of generalized linear model, or break it?
This page describes the following iteratively reweighted linear least-squares (IRLS) method for solving a generalized linear model (GLM):
let $x_1=0$
for $j=1,2,...$ do
linear ...
- 365
1
vote
0
answers
473
views
Negative Value Gamma GLM Inverse Link
I want to calculate a log-likelihood score for a gamma glm with inverse link function. The score will be used to find optimal parameters ($\beta$) for the model. I'm fixing the shape parameter and ...
- 131
1
vote
1
answer
79
views
Log-likelihood using the link identity for poisson?
I understood the Log-likelihood using the link “log” for poisson, λ=exp(α+βx). But I can’t get the Log-likelihood in the case of “identity”, λ=α+βx. How do I get it?. The example is the following data....
- 227