Akaike information criterion
The Akaike information criterion (AIC) is an estimator of the relative quality of statistical models for a given set of data. Given a collection of models for the data, AIC estimates the quality of each model, relative to each of the other models. Thus, AIC provides a means for model selection.
AIC is founded on information theory. When a statistical model is used to represent the process that generated the data, the model will almost never be exact; so some information will be lost by using the model to represent the process. AIC estimates the relative information lost by a given model: the less information a model loses, the higher the quality of that model. (In making an estimate of the information lost, AIC deals with the trade-off between the goodness of fit of the model and the simplicity of the model.)
The Akaike information criterion is named after the statistician Hirotugu Akaike, who formulated it. It now forms the basis of a paradigm for the foundations of statistics; as well, it is widely used for statistical inference.
Suppose that we have a statistical model of some data. Let k be the number of estimated parameters in the model. Let be the maximum value of the likelihood function for the model. Then the AIC value of the model is the following.
Given a set of candidate models for the data, the preferred model is the one with the minimum AIC value. Thus, AIC rewards goodness of fit (as assessed by the likelihood function), but it also includes a penalty that is an increasing function of the number of estimated parameters. The penalty discourages overfitting, because increasing the number of parameters in the model almost always improves the goodness of the fit.
AIC is founded in information theory. Suppose that the data is generated by some unknown process f. We consider two candidate models to represent f: g1 and g2. If we knew f, then we could find the information lost from using g1 to represent f by calculating the Kullback–Leibler divergence, DKL(f ‖ g1); similarly, the information lost from using g2 to represent f could be found by calculating DKL(f ‖ g2). We would then choose the candidate model that minimized the information loss.
We cannot choose with certainty, because we do not know f. Akaike (1974) showed, however, that we can estimate, via AIC, how much more (or less) information is lost by g1 than by g2. The estimate, though, is only valid asymptotically; if the number of data points is small, then some correction is often necessary (see AICc, below).
Note that AIC does not provide a test of a model in the sense of testing a null hypothesis. It tells nothing about the absolute quality of a model, only the quality relative to other models. Thus, if all the candidate models fit poorly, AIC will not give any warning of that.
How to apply AIC in practice
To apply AIC in practice, we start with a set of candidate models, and then find the models' corresponding AIC values. There will almost always be information lost due to using a candidate model to represent the "true model" (i.e. the process that generated the data). We wish to select, from among the candidate models, the model that minimizes the information loss. We cannot choose with certainty, but we can minimize the estimated information loss.
Suppose that there are R candidate models. Denote the AIC values of those models by AIC1, AIC2, AIC3, …, AICR. Let AICmin be the minimum of those values. Then the quantity exp((AICmin − AICi)/2) can be interpreted as being proportional to the probability that the ith model minimizes the (estimated) information loss.
As an example, suppose that there are three candidate models, whose AIC values are 100, 102, and 110. Then the second model is exp((100 − 102)/2) = 0.368 times as probable as the first model to minimize the information loss. Similarly, the third model is exp((100 − 110)/2) = 0.007 times as probable as the first model to minimize the information loss.
In this example, we would omit the third model from further consideration. We then have three options: (1) gather more data, in the hope that this will allow clearly distinguishing between the first two models; (2) simply conclude that the data is insufficient to support selecting one model from among the first two; (3) take a weighted average of the first two models, with weights proportional to 1 and 0.368, respectively, and then do statistical inference based on the weighted multimodel.
The quantity exp((AICmin − AICi)/2) is known as the relative likelihood of model i. It is closely related to the likelihood ratio used in the likelihood-ratio test. Indeed, if all the models in the candidate set have the same number of parameters, then using AIC might at first appear to be very similar to using the likelihood-ratio test. There are, however, important distinctions. In particular, the likelihood-ratio test is valid only for nested models, whereas AIC (and AICc) has no such restriction.
When the sample size is small, there is a substantial probability that AIC will select models that have too many parameters, i.e. that AIC will overfit. To address such potential overfitting, AICc was developed: AICc is AIC with a correction for small sample sizes.
The formula for AICc depends upon the statistical model. Assuming that the model is univariate, is linear in its parameters, and has normally-distributed residuals (conditional upon regressors), then the formula for AICc is as follows.
—where n denotes the sample size and k denotes the number of parameters. Thus, AICc is essentially AIC with an extra penalty term for the number of parameters. Note that as n → ∞, the extra penalty term converges to 0, and thus AICc converges to AIC.
If the assumption that the model is univariate and linear with normal residuals does not hold, then the formula for AICc will generally be different from the formula above. For some models, the formula can be difficult to determine. For every model that has AICc available, though, the formula for AICc is given by AIC plus terms that includes both k and k2. In comparison, the formula for AIC includes k but not k2. In other words, AIC is a first-order estimate (of the information loss), whereas AICc is a second-order estimate.
Further discussion of the formula, with examples of other assumptions, is given by Burnham & Anderson (2002, ch. 7) and by Konishi & Kitagawa (2008, ch. 7–8). In particular, with other assumptions, bootstrap estimation of the formula is often feasible.
To summarize, AICc has the advantage of tending to be more accurate than AIC (especially for small samples), but AICc also has the disadvantage of sometimes being much more difficult to compute than AIC. Note that if all the candidate models have the same k and the same formula for AICc, then AICc and AIC will give identical (relative) valuations; hence, there will be no disadvantage in using AIC, instead of AICc. Furthermore, if n is many times larger than k2, then the extra penalty term will be negligible; hence, the disadvantage in using AIC, instead of AICc, will be negligible.
The Akaike information criterion was formulated by the statistician Hirotugu Akaike; it was originally named "an information criterion". It was first announced by Akaike at a 1971 symposium, the proceedings of which were published in 1973. The 1973 publication, though, was only an informal presentation of the concepts. The first formal publication was a 1974 paper by Akaike. As of October 2014, the 1974 paper had received more than 14000 citations in the Web of Science: making it the 73rd most-cited research paper of all time.
Nowadays, AIC has become common enough that it is often used without citing Akaike's 1974 paper. Indeed, there are over 150,000 scholarly articles/books that use AIC (as assessed by Google Scholar).
The initial derivation of AIC relied upon some strong assumptions. Takeuchi (1976) showed that the assumptions could be made much weaker. Takeuchi's work, however, was in Japanese and was not widely known outside Japan for many years.
AICc was originally proposed for linear regression (only) by Sugiura (1978). That instigated the work of Hurvich & Tsai (1989), and several further papers by the same authors, which extended the situations in which AICc could be applied.
The first general exposition of the information-theoretic approach was the volume by Burnham & Anderson (2002). It includes an English presentation of the work of Takeuchi. The volume led to far greater use of AIC, and it now has more than 42000 citations on Google Scholar.
Akaike called his approach an "entropy maximization principle", because the approach is founded on the concept of entropy in information theory. Indeed, minimizing AIC in a statistical model is effectively equivalent to maximizing entropy in a thermodynamic system; in other words, the information-theoretic approach in statistics is essentially applying the Second Law of Thermodynamics. As such, AIC has roots in the work of Ludwig Boltzmann on entropy. For more on these issues, see Akaike (1985) and Burnham & Anderson (2002, ch. 2).
A statistical model must fit all the data points. Thus, a straight line, on its own, is not a model of the data, unless all the data points lie exactly on the line. We can, however, choose a model that is "a straight line plus noise"; such a model might be formally described thus: yi = b0 + b1xi + εi. Here, the εi are the residuals from the straight line fit. If the εi are assumed to be i.i.d. Gaussian (with zero mean), then the model has three parameters: b0, b1, and the variance of the Gaussian distributions. Thus, when calculating the AIC value of this model, we should use k=3. More generally, for any least squares model with i.i.d. Gaussian residuals, the variance of the residuals' distributions should be counted as one of the parameters.
As another example, consider a first-order autoregressive model, defined by xi = c + φxi−1 + εi, with the εi being i.i.d. Gaussian (with zero mean). For this model, there are three parameters: c, φ, and the variance of the εi. More generally, a pth-order autoregressive model has p + 2 parameters. (If, however, c is not estimated from the data, but instead given in advance, then there are only p + 1 parameters.)
The AIC values of the candidate models must all be computed with the same data set. Sometimes, though, we might want to compare a model of the response variable, y, with a model of the logarithm of the response variable, log(y). More generally, we might want to compare a model of the data with a model of transformed data. Following is an illustration of how to deal with data transforms (adapted from Burnham & Anderson (2002, §2.11.3): "Investigators should be sure that all hypotheses are modeled using the same response variable").
Suppose that we want to compare two models: one with a normal distribution of y and one with a normal distribution of log(y). We should not directly compare the AIC values of the two models. Instead, we should transform the normal cumulative distribution function to first take the logarithm of y. To do that, we need to perform the relevant integration by substitution: thus, we need to multiply by the derivative of the (natural) logarithm function, which is 1/y. Hence, the transformed distribution has the following probability density function:
—which is the probability density function for the log-normal distribution. We then compare the AIC value of the normal model against the AIC value of the log-normal model.
Some statistical software will report the value of AIC or the maximum value of the log-likelihood function, but the reported values are not always correct. Typically, any incorrectness is due to a constant in the log-likelihood function being omitted. For example, the log-likelihood function for n independent identical normal distributions is
—this is the function that is maximized, when obtaining the value of AIC. Some software, however, omits the constant term (n/2) ln(2π), and so reports erroneous values for the log-likelihood maximum—and thus for AIC. Such errors do not matter for AIC-based comparisons, if all the models have their residuals as normally-distributed: because then the errors cancel out. In general, however, the constant term needs to be included in the log-likelihood function. Hence, before using software to calculate AIC, it is generally good practice to run some simple tests on the software, to ensure that the function values are correct.
Comparisons with other model selection methods
Comparison with BIC
The formula for the Bayesian information criterion (BIC) is similar to the formula for AIC, but with a different penalty for the number of parameters. With AIC the penalty is 2k, whereas with BIC the penalty is ln(n) k.
A comparison of AIC/AICc and BIC is given by Burnham & Anderson (2002, §6.3-6.4), with follow-up remarks by Burnham & Anderson (2004). The authors show that AIC/AICc can be derived in the same Bayesian framework as BIC, just by using different prior probabilities. In the Bayesian derivation of BIC, though, each candidate model has a prior probability of 1/R (where R is the number of candidate models); such a derivation is "not sensible", because the prior should be a decreasing function of k. Additionally, the authors present a few simulation studies that suggest AICc tends to have practical/performance advantages over BIC.
A point made by several researchers is that AIC and BIC are appropriate for different tasks. In particular, BIC is argued to be appropriate for selecting the "true model" (i.e. the process that generated the data) from the set of candidate models, whereas AIC is not appropriate. To be specific, if the "true model" is in the set of candidates, then BIC will select the "true model" with probability 1, as n → ∞; in contrast, when selection is done via AIC, the probability can be less than 1. Proponents of AIC argue that this issue is negligible, because the "true model" is virtually never in the candidate set. Indeed, it is a common aphorism in statistics that "all models are wrong"; hence the "true model" (i.e. reality) cannot be in the candidate set.
Another comparison of AIC and BIC is given by Vrieze (2012). Vrieze presents a simulation study—which allows the "true model" to be in the candidate set (unlike with virtually all real data). The simulation study demonstrates, in particular, that AIC sometimes selects a much better model than BIC even when the "true model" is in the candidate set. The reason is that, for finite n, BIC can have a substantial risk of selecting a very bad model from the candidate set. This reason can arise even when n is much larger than k2. With AIC, the risk of selecting a very bad model is minimized.
If the "true model" is not in the candidate set, then the most that we can hope to do is select the model that best approximates the "true model". AIC is appropriate for finding the best approximating model, under certain assumptions. (Those assumptions include, in particular, that the approximating is done with regard to information loss.)
Comparison of AIC and BIC in the context of regression is given by Yang (2005). In regression, AIC is asymptotically optimal for selecting the model with the least mean squared error, under the assumption that the "true model" is not in the candidate set. BIC is not asymptotically optimal under the assumption. Yang additionally shows that the rate at which AIC converges to the optimum is, in a certain sense, the best possible.
Comparison with cross-validation
Comparison with least squares
Sometimes, each candidate model assumes that the residuals are distributed according to independent identical normal distributions (with zero mean). That gives rise to least squares model fitting.
With least squares fitting, the maximum likelihood estimate for the variance of a model's residuals distributions is , where is the residual sum of squares: . Then, the maximum value of a model's log-likelihood function is
—where C1 is a constant independent of the model, and dependent only on the particular data points, i.e. it does not change if the data does not change.
That gives AIC = 2k + n ln(RSS/n) − 2C1 = 2k + n ln(RSS) + C2. Because only differences in AIC are meaningful, the constant C2 can be ignored, which allows us to conveniently take AIC = 2k + n ln(RSS) for model comparisons. Note that if all the models have the same k, then selecting the model with minimum AIC is equivalent to selecting the model with minimum RSS—which is the usual objective of model selection based on least squares.
Comparison with Mallows's Cp
Mallows's Cp is equivalent to AIC in the case of (Gaussian) linear regression.
- Burnham & Anderson 2002, §2.2
- Akaike 1974
- Burnham & Anderson 2002, §2.9.1, §6.4.5
- Burnham & Anderson 2002
- Burnham & Anderson 2002, §2.12.4
- Murtaugh 2014
- McQuarrie & Tsai 1998
- Claeskens & Hjort 2008, §8.3
- Giraud 2015, §2.9.1
- Cavanaugh 1997
- Burnham & Anderson 2002, §2.4
- Burnham & Anderson 2004
- Burnham & Anderson 2002, §7.4
- Akaike 1973
- deLeeuw 1992
- Van Noordon R., Maher B., Nuzzo R. (2014), "The top 100 papers", Nature, 514.
- Sources containing both "Akaike" and "AIC"—at Google Scholar.
- Burnham & Anderson 2002, p. 63
- Burnham & Anderson 2002, p. 82
- Burnham & Anderson 2002, §6.3-6.4
- Vrieze 2012
- Aho et al. 2014
- Stone 1977
- Fang 2011
- Burnham & Anderson 2002, p. 63
- Boisbunon et al. 2014
- Aho, K.; Derryberry, D.; Peterson, T. (2014), "Model selection for ecologists: the worldviews of AIC and BIC", Ecology, 95: 631–636, doi:10.1890/13-1452.1 .
- Akaike, H. (1973), "Information theory and an extension of the maximum likelihood principle", in Petrov, B. N.; Csáki, F., 2nd International Symposium on Information Theory, Tsahkadsor, Armenia, USSR, September 2-8, 1971, Budapest: Akadémiai Kiadó, pp. 267–281 .
- Akaike, H. (1974), "A new look at the statistical model identification", IEEE Transactions on Automatic Control, 19 (6): 716–723, doi:10.1109/TAC.1974.1100705, MR 0423716 .
- Akaike, H. (1985), "Prediction and entropy", in Atkinson, A. C.; Fienberg, S. E., A Celebration of Statistics, Springer, pp. 1–24 .
- Boisbunon, A.; Canu, S.; Fourdrinier, D.; Strawderman, W.; Wells, M. T. (2014), "Akaike's Information Criterion, Cp and estimators of loss for elliptically symmetric distributions", International Statistical Review, 82: 422–439, doi:10.1111/insr.12052 .
- Burnham, K. P.; Anderson, D. R. (2002), Model Selection and Multimodel Inference: A practical information-theoretic approach (2nd ed.), Springer-Verlag .
- Burnham, K. P.; Anderson, D. R. (2004), "Multimodel inference: understanding AIC and BIC in Model Selection" (PDF), Sociological Methods & Research, 33: 261–304, doi:10.1177/0049124104268644 .
- Cavanaugh, J. E. (1997), "Unifying the derivations of the Akaike and corrected Akaike information criteria", Statistics & Probability Letters, 31: 201–208, doi:10.1016/s0167-7152(96)00128-9 .
- Claeskens, G.; Hjort, N. L. (2008), Model Selection and Model Averaging, Cambridge University Press . [Note: the AIC defined by Claeskens & Hjort is the negative of the standard definition—as originally given by Akaike and followed by other authors.]
- deLeeuw, J. (1992), "Introduction to Akaike (1973) information theory and an extension of the maximum likelihood principle" (PDF), in Kotz, S.; Johnson, N. L., Breakthroughs in Statistics I, Springer, pp. 599–609 .
- Fang, Yixin (2011), "Asymptotic equivalence between cross-validations and Akaike Information Criteria in mixed-effects models" (PDF), Journal of Data Science, 9: 15–21 .
- Giraud, C. (2015), Introduction to High-Dimensional Statistics, CRC Press .
- Hurvich, C. M.; Tsai, C.-L. (1989), "Regression and time series model selection in small samples", Biometrika, 76: 297–307, doi:10.1093/biomet/76.2.297 .
- Konishi, S.; Kitagawa, G. (2008), Information Criteria and Statistical Modeling, Springer .
- McQuarrie, A. D. R.; Tsai, C.-L. (1998), Regression and Time Series Model Selection, World Scientific, ISBN 981-02-3242-X .
- Murtaugh, P. A. (2014), "In defense of P values", Ecology, 95: 611–617, doi:10.1890/13-0590.1 .
- Stone, M. (1977), "An asymptotic equivalence of choice of model by cross-validation and Akaike's criterion", Journal of the Royal Statistical Society, Series B, 39 (1): 44–47, JSTOR 2984877 .
- Sugiura, N. (1978), "Further analysis of the data by Akaike's information criterion and the finite corrections", Communications in Statistics - Theory and Methods, 7: 13–26, doi:10.1080/03610927808827599 .
- Takeuchi, K. (1976), " " [Distribution of informational statistics and a criterion of model fitting], Suri-Kagaku [Mathematical Sciences] (in Japanese), 153: 12–18 .
- Vrieze, S. I. (2012), "Model selection and psychological theory: a discussion of the differences between the Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC)", Psychological Methods, 17: 228–243, doi:10.1037/a0027127, PMC 3366160, PMID 22309957 .
- Yang, Y. (2005), "Can the strengths of AIC and BIC be shared?", Biometrika, 92: 937–950, doi:10.1093/biomet/92.4.937 .
- Akaike, H. (21 December 1981), "This Week's Citation Classic" (PDF), Current Contents Engineering, Technology, and Applied Sciences, 12 (51): 42 [Hirotogu Akaike comments on how he arrived at AIC]
- Anderson, D. R. (2008), Model Based Inference in the Life Sciences, Springer
- Burnham, K. P.; Anderson, D. R.; Huyvaert, K. P. (2011), "AIC model selection and multimodel inference in behavioral ecology" (PDF), Behavioral Ecology and Sociobiology, 65: 23–35, doi:10.1007/s00265-010-1029-6
- Ing, C.-K.; Wei, C.-Z. (2005), "Order selection for same-realization predictions in autoregressive processes", Annals of Statistics, 33: 2423–2474, arXiv:math/0602326, doi:10.1214/009053605000000525
- Pan, W. (2001), "Akaike's Information Criterion in generalized estimating equations" (PDF), Biometrics, 57: 120–125, doi:10.1111/j.0006-341X.2001.00120.x
- Parzen, E.; Tanabe, K.; Kitagawa, G., eds. (1998), Selected Papers of Hirotugu Akaike, Springer, doi:10.1007/978-1-4612-1694-0
- Saefken, B.; Kneib, T.; van Waveren, C.-S.; Greven, S. (2014), "A unifying approach to the estimation of the conditional Akaike information in generalized linear mixed models", Electronic Journal of Statistics, 8: 201–225, doi:10.1214/14-EJS881