Stack Exchange Network

Stack Exchange network consists of 174 Q&A communities including Stack Overflow, the largest, most trusted online community for developers to learn, share their knowledge, and build their careers.

Visit Stack Exchange

Questions tagged [aic]

AIC stands for the Akaike Information Criterion, which is one technique used to select the best model from a class of models using a penalized likelihood. A smaller AIC implies a better model.

0
votes
0answers
17 views

Relation between the AIC and the Kullback-Leibler Divergence

I am searching a formal derivation of the Akaike Information Criterion from the Kullback-Leibler Divergence. Can you show me one, or point me toward a book/article in which this is done? Here I set ...
1
vote
1answer
32 views

Extremely large difference in AICs between two models

I am currently fitting a mixed model where I analye longitudinal trends in migration between country pairs (68335 observations nested in 6442 groups). One of the first questions I wanted to have ...
0
votes
0answers
22 views

Generalising AIC results over multiple samples

This is slightly related to my previous question (AIC Calculation using log likelihood) Though, I think now I am actually clear as to what I am asking. I am modelling activity of cells, I have data ...
0
votes
0answers
15 views

Selection optimum polynomial fit

I have fitted polynomial model of orders 1-4. I have three predictors with 7 levels and my response is 400 values from 0.6-0.9, which seems to be bad for information criterions. I am interested in ...
3
votes
0answers
67 views

Is the use of loglik or AIC to compare logit/probit/cloglog models valid?

I would like to know whether I can use AIC, or if the models have the same number of predictors, the log-likelihood, to compare logit vs probit vs cloglog models (fitted for instance with glmer or ...
0
votes
1answer
68 views

AIC Calculation using log likelihood

I have a dataset that has 40 experimental observations of cells' activity, $n=40$, I tested several models using each of these samples. The model can only explain one cell at a time due to variability ...
3
votes
0answers
35 views

Standardizing qualitative variables in R to perform glm's, glm.nb's and lm's [closed]

I want to standardize the variables of a biological dataset. I need to run glm's, glm.nb's and lm's using different response variables but the same explanatory variables. The dataset contains counts ...
0
votes
2answers
27 views

Calculating the relative likelihood with AIC values

I'm using AIC for model selection, and would like to use a relative likelihood measure to quantify how many times a model with minimum AIC (AICmin) fits better than the alternative (with AICi). For ...
1
vote
1answer
38 views

Understanding Intuition for ETS Damping Selection via AIC/BIC

I'm trying to understand how ETS selects whether to use a damped model via information criteria (I'm not sure which of AIC, AICc or BIC are used). I have a time series and I'm comparing two ETS ...
1
vote
0answers
36 views

AIC formula in R vs Python [closed]

I have been trying to calculate a GLM's AIC both in python (package Statsmodels) and R (native glm function). For exactly the same model I get two different AIC estimates. The formula for AIC is: -...
2
votes
0answers
39 views

When to disregard AIC as a criterion in model selection

I have the following problem: I'm working on a dataset and it looks completely quadratic. A quadratic regression fits the data really good. However, when using piecewise linear functions I get a lower ...
2
votes
1answer
66 views

Selecting between OLS regression and ARIMA for time series, why AIC or BIC for ARIMA is much larger in my data?

My data set is quarterly time seires data (around 140 data points). Method 1: simple OLS regression with 5-6 exogenous variables, which are drivers of the dependent variable. None of the explanatory ...
3
votes
2answers
47 views

Better AIC but worse cross validation error rate

I learn that AIC is usually used for assessing goodness of fit of a model and the criterion takes into account both goodness of fit and number of parameters used so that it could regulates the issue ...
1
vote
1answer
32 views

How many candidate models to include in AIC model selection?

Is there a rule of thumb, perhaps related to sample size, for how many models to include in AIC model selection? Too many may seem like fishing while too few would be insufficient. I'm familiar with ...
0
votes
0answers
25 views

Fitting an ARMA-GARCH using AIC

I am trying to fit an ARFIMA(p,d,q)-GARCH(1,1) model to an asset returns time series. I start with an ARFIMA(0,0,0)-GARCH(1,1). The diagnostics tests like persistence requirement, Ljung Box test for ...
0
votes
0answers
28 views

Handling collinearity in GAM

I am interesting if a variable x1 is of importance for the outcome y. To investigate this I am trying to fit a model A1 <- gam(y~ s(x1) + s(x2) + x3 + x4 + x5 + x6)) where x1 and x2 are ...
15
votes
1answer
508 views

Stepwise AIC - Does there exist controversy surrounding this topic?

I've read countless posts on this site that are incredibly against the use of stepwise selection of variables using any sort of criterion whether it be p-values based, AIC, BIC, etc. I understand why ...
0
votes
1answer
24 views

AIC calculated in lm(y~1) and stepwise selection in R

http://www.stat.wisc.edu/courses/st333-larget/aic.pdf The AIC calculated with the model lm(SAT~1) was 560.4736, but the AIC calculated with stepwise selection starting with lm(SAT~1) was 419.42. May ...
0
votes
0answers
32 views

Shrinkage methods - are they any good for statistical inference or should they be used for prediction goals only?

I am working on my master thesis with a goal to find regressors which influence companies' decisions on how to pay for a target in acquisitions (cash, stock or a mix of both). I have 13 regressors to ...
1
vote
0answers
43 views

AIC based model selection, hyperparameter optimization and in-sample prediction

I'm using AIC to perform model selection along with hyperparameters optimization. The exact setup is the following: I have two input variables (A and B), and a single target variable. All variables ...
0
votes
0answers
11 views

Information criterion when model could be mis-specified and data is dependent

Common information criteria (AIC, BIC, etc) require the user to specify the likelihood function, while in practice rarely the user has the luxury to know the correct likelihood function. In the case ...
4
votes
1answer
74 views

How is the Akaike information criterion (AIC) affected by sample size?

I am evaluating several logistic regression models predicting college student retention. I am using some basic and well-established predictors, such as high school GPA and SAT scores. I understand ...
0
votes
0answers
31 views

Aikaike Information Criterion: derivation in original paper

I have been reading AIC paper 'Information theory and an extension of the maximum likelihood principle' by Akaike (1974). I have been able to understand up to the third section of the paper, but I am ...
0
votes
0answers
55 views

Convince me that AIC can't be used to compare models with different sample sizes

Conventional statistical wisdom says we cannot compare AIC (or other information criteria predictive sample statistics) when the sample sizes for the compared models are different (See here for ...
0
votes
0answers
9 views

DIC/MDIC is negative. Is this a red-flag or normal?

I have fit a few mixture models of multivariate normals. When calculating the MDIC (modified DIC) to compare the models, they come up as negative. In contrast, all examples in my lecture notes are ...
0
votes
1answer
29 views

How can we calculate AIC from a negative binomial GLMM?

Our problem here described is to calculate AIC from a GLMM negbin. Our data compose by 2 Categorical variables (Yes/Not), 3 Numerical variables and our random factor, all without any NA. We want to ...
1
vote
0answers
28 views

Linear regression AIC and Randomisation Test

The problem is that I used AIC as the criterion for model selection and that gives me a model with 3 parameters (the model has the lowest AIC). $$y = \beta_0 + \beta_1X_1+\beta_2X_2+\beta_3X_3$$ ...
1
vote
0answers
34 views

Effect sizes for model averaging

The goal is to find if one factor is stronger than the other in the models I have considered. I am using the information-theoretic approach. Since $n/K>40$, I am using AIC. Firstly the model is ...
0
votes
0answers
54 views

How to optimise Linear Predictive Coding model using AIC/BIC

I am currently using an LPC function in Python that takes a signal x and an order and returns the solution a, the prediction error e, and the reflection coefficients k. The aim is to fit a filter to ...
1
vote
1answer
32 views

Why do we use a criterion like AIC for Copula model selection?

If we look at the AIC formula: AIC = -2*log(ML) + 2k where k is the number of parameters in the model and is considered as the 'penalizing term' for complexity or over-fitting. Does this ...
0
votes
0answers
29 views

Can you use AIC to compare OLS and SMA regressions?

I've come across SMA (Standard Major Axis) regression recently and was wondering if it would be appropriate to compare SMA and OLS models with AIC? When I use the aicc function from the package bbmle, ...
0
votes
0answers
40 views

EM algorithm and AIC criteria

I am using EM algorithm to estimate the model parameters. EM-algorithm iterates until the loglikelihood is converged. After that, I need to compute AIC criteria. As known, AIC is a loglikelihood ...
1
vote
0answers
18 views

AIC and BIC and number of quantization level

I want to test how many quantization levels (discretizing levels) are the best for the given data(time series) set I have. Therefore I am applying different levels of binning (like discretisize data ...
1
vote
0answers
68 views

Proof of AIC criterion

I was recently read the book "Elements Of Statistical Learning" by Hastie et.al. In chapter $7$, AIC criterion is definite as follows: $$-2.E[log Pr_\hat\theta(Y)]\approx -\frac{2}{N}.E[loglik]+2\...
2
votes
0answers
82 views

Time series analysis VAR model: AIC and BIC test criteria

Consider two variables. Imagine you want to analyse the effects of the lags of variable A on variable B. The possiblity you see an effect of variable A on B is reasonable, but there is absolutely no ...
4
votes
2answers
273 views

What is the actual significance of a difference in AIC or BIC values?

Usually, when a difference of a statistic is discussed, that discussion is presented in the context of a significance of that difference. When self-entropy, i.e., information content, is examined, ...
3
votes
1answer
191 views

Can one give an example(s) of when non-nested AIC model comparison is not useful for model selection?

Note: The question here is not the same as this one. Indeed, as an answer to that question the answer below was closed as unrelated, together with the suggestion (credit @gung) to ask a separate ...
6
votes
1answer
311 views

Elbow Test using AIC/BIC for identifying number of clusters using GMM

How to select number of clusters using GMM when the elbow test (AIC/BIC vs n_components) results in a graph like this?
0
votes
1answer
28 views

How many parameters should I report in an AIC table for a tobit model?

I am considering using a tobit model in R to estimate velocity of a number of animals moving north across a landscape. I am working with simulated data, so when an ...
0
votes
0answers
53 views

Assuming heteroscedasticity produces terrible QQ plot

I created a model using the lm function and obtained a model with an AIC of 390516.1 and the following QQ-plot: However, the Breusch-Pagan Test returned a p-value ...
1
vote
0answers
34 views

AIC and BIC in Latent class analysis

I am using the Latent Class Analysis feature available in Stata 15. The two statistical criterions gave me different indications: $AIC$ suggests me to use 6 classes, instead $BIC$ suggests to use 5 ...
0
votes
0answers
33 views

How do you “sell” having too many variables? - all of them are important, but your data does not allow that many

Scenario: You run a binary logistic regression. You want to find out if it is possible to build a model to predict a higher likelihood of the dependent variable to be 1, based on the data you can get ...
1
vote
0answers
25 views

What is the correct multiple regression method to answer my hypotheses?

I have a dataset in which I'm looking to explore the relationships between nine input variables (X1 - X9) and five outputs (Y1 - Y5). For each output, I have a hypothesis based on the previous ...
0
votes
0answers
21 views

How to interpret the output from an all-subsets regression?

I have a dataset in which I'm looking to explore the relationships between nine input variables (X1 - X9) and five outputs (Y1 - Y5). For some of the outputs I hypothesise that certain inputs will be ...
1
vote
1answer
31 views

Akaike information criterion. AICs analysis ( sample size =5, No of parameters (not counting the error variance) = 3), in result AICc = infinite.

Could you please recommend another formula for AIC correction for small sample size (n/k < 40)
5
votes
1answer
330 views

When is the AIC a good model selection criterion for forecasting and when is it not?

I'm trying to wrap my head around why the AIC and other similar ICs work as proxies for out of sample error when trying to perform automated forecast generation. So I performed an experiment on the ...
0
votes
0answers
32 views

Compare models with robust methds - R

This question is the first part of a larger question that is continued here. I thought that it could be easier to split it into two questions to generate better answer to both and to help further ...
1
vote
0answers
27 views

Simple Global Temperature Models [closed]

I'm a volunteer for 350, a grassroots organization that is fighting against climate change through education and advocacy. We're all volunteers, not climate scientists or statisticians, so I'm looking ...
1
vote
1answer
41 views

Q: Number of AIC parameters (again)

I know very similar questions get asked a lot - but I would like to make sure that I am not missing something here: Suppose we have the simple model: y = mx + c + s*e where e ~ N(0,1) and m, c and ...
2
votes
1answer
73 views

Glmm models -> AIC model selection -> Several models have the same “explanationary power” (similar AIC) -> now what?

So, I have a bunch of models, I'm using AIC for model selection, I don't have the exact numbers in front of me now but let's take for example: Model 1 - AIC = 100 Model 2 - AIC = 101; $\Delta$AIC = 1 ...