Akaike information criterion statistics

By 坂元 慶行, 石黒 真木夫, 北川 源四郎. 1986

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

1. K. A. Triantis
2. P. López‐López
3. M. Forchhammer
4. J. Seoane
5. E. Post
6. E. Solberg
7. T. Buckley
8. G. De’ath

Citations in the corpus

• Statistical and Machine Learning forecasting methods: Concerns and ways forward (2018, background): “…problem since, in contrast to linear statistical methods, where over-fitting can be directly controlled by some information criteria (e.g., the AIC [68]) taking into account the number of parameters utilized, ML methods are nonlinear and training is performed dynamically, meaning that different…”
• Model selection and model averaging in phylogenetics: advantages of akaike information criterion and bayesian approaches over likelihood ratio tests. (2004, methodology): “A different approach to model selection is the Akaike Information Criterion (AIC) (Akaike, 1973, 1974; and see Sakamoto et al., 1986).”
• Species richness, environmental heterogeneity and area: a case study based on land snails in Skyros archipelago (Aegean Sea, Greece) (2005, background): “The difference in AIC values (DAIC 1⁄4 AICi ) AICj) indicates the best model, since the model with the lowest AIC value is considered to be better (Sakamoto et al., 1986; Burnham & Anderson, 2002).”
• A statistical analysis and comparison of historical earthquake and tsunami disasters in Japan and Indonesia (2014, background): “However, since earthquakes and tsunamis are rare events, unpredictable and stochastic natural phenomena as described in Section 3, in terms of p-0 and n-1, taking the limit so that λ1⁄4np, we know we can approximate the probability of the Negative Binomial by the Poisson distribution [22].”
• Predicting the distribution of four species of raptors (Aves: Accipitridae) in southern Spain: statistical models work better than existing maps (2004, methodology): “The step.gam procedure uses a stepwise search to select the best model in terms of Akaike’s information criterion (AIC) that takes into account both the information explained by the model and its complexity (the lesser the AIC, the better the model Sakamoto et al., 1986).”
• BREEDING DISPERSAL IN FEMALE NORTH AMERICAN RED SQUIRRELS (2000, methodology): “We selected the model that best fit the data using a forward stepwise procedure and the Akaike Information Criterion (AIC) (Sakamoto et al. 1986, Burnham and Anderson 1992).”
• Coping With Customer Complaints (2006, result): “Although the fit of the hypothesized model (AIC = 162.51) was slightly better than the competing model (AIC = 163.23), the difference is not significant using guidelines suggested by Sakamoto, Ishiguro, and Kitagawa (1986).”
• Higher genetic diversity in introduced than in native populations of the mussel Mytella charruana: evidence of population admixture at introduction sites (2009, unknown): “The bestfit model of evolution was selected for COI using AIC (Akaike, 1973, 1974; Sakamoto et al., 1986) implemented in ModelT a b le 1 E st im at es o f M yt el la sp p .”
• How many cosmological parameters (2004, methodology): “The standard approach thus far has been to first choose the set of parameters to be varied on a fairly ad hoc basis, and then use a likelihood method to find the best-fitting model and confidence ranges for those parameters.”
• Applied Linear Regression (2004, background): “10.2 variable selection for discovery 239 every candidate subset, aic is given for linear regression by Sakamoto et al. (1986), AIC /= +n n pplog( )RSS C C2 (10.6) Small values of aic are preferred, so better candidate sets will have smaller RSS and a smaller number of terms pC .”
• A model for the species–area–habitat relationship (2003, methodology): “Comparing two models, the one with the lowest value of AIC is considered to be better.”
• The Realized Yield Effect of Genetically Engineered Crops: U.S. Maize and Soybean (2013, unknown): “Although the Akaike information criterion (Sakamoto et al., 1986) that could be used in this case (note that the two models are not nested) would favor the shift model, the difference is small and it seems unwise to rely on this model selection procedure because the two models have distinctly…”
• Effects of density-dependence and climate on the dynamics of a Svalbard reindeer population (2001, methodology): “The most parsimonious model was selected by the corrected Akaike information criterion, AICc (Hurvich and Tsai 1989), the model with lowest AICc being considered most parsimonious (Sakamoto et al. 1986).”
• Modelling breeding habitat preferences of Bonelli’s eagle (Hieraaetus fasciatus) in relation to topography, disturbance, climate and land use at different spatial scales (2005, methodology): “The smaller the AICc, the better the model (Sakamoto et al. 1986). aSee text for details Receiver operating characteristic (ROC) plot was computed to assess the power of the logistic models (Gibson et al. 2004; Osborne et al. 2001; Pearce and Ferrier 2000b).”
• Local Models for Forest Canopy Cover with Beta Regression (2007, methodology): “However, if the second and third order coefficients were not statistically significant and increased the model AIC (low AIC indicates a good balance between model fit and the number of parameters), the linear relation was used instead.”
• Genetic Improvement of U.S. Soybean in Maturity Groups II, III, and IV (2014, background): “The AIC rewards models with a better coefficient of determination (R2) fit, but penalizes models that require more parameters to achieve the better fit (Sakamoto et al., 1986; Posada and Buckley, 2004).”