Performance of several variable‐selection methods applied to real ecological data

This paper introduces a SAS/IML program to select among the multivariate model candidates based on a few well-known multivariate model selection criteria. Stepwise regression and all-possible-regression are considered. The program is user friendly and requires the user to paste or read the data at the beginning of the module, include the names of the dependent and independent variables (the y's and the x's), and then run the module. The program produces the multivariate candidate models based on the following criteria: Forward Selection, Forward Stepwise Regression, Backward Elimination, Mean Square Error, Coefficient of Multiple Determination, Adjusted Coefficient of Multiple Determination, Akaike's Information Criterion, the Corrected Form of Akaike's Information Criterion, Hannan and Quinn Information Criterion, the Corrected Form of Hannan and Quinn (HQc) Information Criterion, Schwarz's Criterion, and Mallow's PC. The output also constitutes detailed as well as summarized results.

In this paper, we are interested in investigating the causal relationships among futures sugar prices in the Zhengzhou futures exchange market (ZF), the spot sugar prices in Zhengzhou (ZS) and the futures sugar prices in New York futures exchange market (NF). A useful tool called Bayesian network is introduced to analyze the problem. Since there are only three variables in our Bayesian network, the algorithm is straightforward: we display all the 25 possible network structures and adopt certain scoring metrics to evaluate them. We applied five different scoring metrics in total. Firstly, for each metric, we obtained 24 scores, each calculated from one of the 24 possible structures i.e. a Directed Acyclic Graph (DAG). Then we eliminated the network structure which represents the independence of the three variables according to our prior knowledge concerning the futures sugar market. After that, the optimal network structure which implies the causal relationships was selected according to the corresponding scoring metric. Finally, after comparing the results from different scoring metrics, we obtained the relatively affirmative conclusion that ZS causes ZF from both the Bayesian Dirichlet (BD) metric, Bayesian Dirichlet-Akaike Information Criterion (BD-AIC) metric, Bayesian Dirichlet-Bayesian Information Criterion (BD-BIC) metric and Bayesian Information Criterion (BIC) metric. The conclusions that NF causes ZF and ZF causes ZS from the Akaike Information Criterion (AIC) metric and ZF causes ZS from the BIC metric were useful and significant to our investigation.

SummaryCase-cohort designs are widely used in large cohort studies to reduce the cost associated with covariate measurement. In many such studies the number of covariates is very large, so an efficient variable selection method is necessary. In this paper, we study the properties of a variable selection procedure using the smoothly clipped absolute deviation penalty in a case-cohort design with a diverging number of parameters. We establish the consistency and asymptotic normality of the maximum penalized pseudo-partial-likelihood estimator, and show that the proposed variable selection method is consistent and has an asymptotic oracle property. Simulation studies compare the finite-sample performance of the procedure with tuning parameter selection methods based on the Akaike information criterion and the Bayesian information criterion. We make recommendations for use of the proposed procedures in case-cohort studies, and apply them to the Busselton Health Study.

Time series study is of paramount importance especially in developing countries such as India. The objective of this study is to forecast the number of dengue cases by applying time series models. Furthermore, this study also aims to uncover the correlation of ecological variables with dengue cases using a p value test. The ecological variables considered are the rainfall (with a 2-month lag period) and average relative humidity. An evaluation of host factors such as age and gender is done with the number of cases to determine the most affected strata. Monthly data on dengue cases from 2014 to 2017 are taken from the integrated disease surveillance programme, Government of India, and data on ecological variables, humidity and rainfall are taken from www.Indiastat.com . A 2-month lag period is considered for rainfall. These data were analysed by the autoregressive integrated moving average (ARIMA) and exponential smoothing model (ES). Predictive performance of these techniques was assessed by evaluation parameters such as the Akaike information criterion (AIC), Bayesian information criterion (BIC), mean absolute error and mean absolute scaled error. Applying time series models was useful in the prediction. There were many more males than females according to the data for 2014–2017. The Akaike information criterion (AIC), Bayesian information criterion (BIC) and mean absolute scaled error for ARIMA are 48.6526, 448.333333, 449.30008, 453.08 and 0.507826, respectively. Based on model indices such as the Akaike information criterion (AIC), Bayesian information criterion (BIC), mean absolute error and mean absolute scaled error, the ARIMA performed better than the ES model. Rainfall at the 2-month lag was correlated with the number of cases.

In this study it was aimed to evaluate the effects of various factors such as sample sizes, percentage of misfit items in the test and item quality (item discrimination) on item and model fit in case of misspecification of Q matrix. Data were generated in accordance with DINA model. Q matrix was specified for 4 attributes and 15 items. While data were generated, sample sizes as 1000, 2000, 4000, s and g parameters as low and high discrimination index were manipulated. Three different misspecified Q matrix (overspecified, underspecified and mixed) was developed considering the percentage of misfit items (%20 and %40). In the study, S-X2 was used as item fit statistics. Furthermore absolute (abs(fcor), max(X2)) and relative (-2 log-likelihood, Akaike's information criterion (AIC) and Bayesian information criterion (BIC)) model fit statistics were used. Investigating the results obtained from this simulation study, it was concluded that S-X2 was sufficient to detect misfit items. When the percentage of misfit items in the test was much or Q matrix was both underspecified and overspecified, the correct detection of both abs(fcor) and max(X2) statistics was approximately 1 or 1. In addition, the correct detection rates of both statistics was high under other conditions, too. AIC and BIC were successful to detect model misfit in the cases where the Q matrix underspecified, whereas, they were failed detect model misfit for other cases. It can be said that the performance of BIC was mostly better than other relative model fit statistics to detect model misfit.

Obesity adversely affects health-related quality of life (HRQoL). Most studies have used body mass index (BMI) to measure obesity. Other measures of obesity, such as waist circumference (WC) or waist-to-hip ratio (WHR), may be better predictors of cardiovascular disease and mortality. We, therefore, examined the associations between other anthropometric measures and HRQoL in a multi-ethnic Asian population. In this follow-up study from four previous cross-sectional surveys, HRQoL was measured, at follow-up, using the Short-Form 36 version 2 (SF-36v2) questionnaire. Linear regression was used to assess the relationship between anthropometric measures [BMI, WC, waist residuals (WR) (generated by regressing WC on BMI), WHR, waist-to-height ratio (WHtR) and height (Ht)] and HRQoL. We compared the models' R2, Akaike's information criteria (AIC), and Schwarz Bayesian information criteria (BIC) from the different models. Among 4,981 subjects, 47.6% were men aged 50.6 ± 12.2 and women aged 49.3 ± 11.6 years. All gender-specific anthropometric measures were significantly correlated with BMI, except WR. After adjusting for known determinants of HRQoL, we found significant associations between BMI, WC and WHtR with SF-36v2 Physical Component Summary (PCS) scores in women but not men. In contrast, after adjusting for known determinants of HRQoL, WR and WHR were significantly associated with SF-36v2 Mental component summary (MCS) scores in men, but not women. R², AIC and BIC were similar for all anthropometric measures in the final model. The associations between measures of central obesity and HRQoL differed between men and women. In women, associations were seen with SF-36v2 PCS, but measures of central obesity did not have significant associations with HRQoL after controlling for BMI. In men, an association between WC and SF-36v2 MCS was statistically significant independent of BMI. These gender differences require further investigation.

This study was conducted to determine the live weight model of the broiler chicks by using the most appropriate mathematical growth curves. Live weights were used in broiler chicks grown for 0-6 weeks. Logistics, Gompertz, Weibull, Hossfeld and Von Bertalanffy models and multivariate adaptive regression splines (MARS) data mining algorithm were used to define the live weights of the chickens. In the comparison of the models, the determination coefficient (R2), mean square error (MSE), Akaike's Information Criterion (AIC) and Schwarz Bayesian Information Criterion (BIC) values were used. As a result of the study, it is seen that Gompertz model is the best model to define live weight of the broilers in the Gompertz model, R2, MSE, RMSE, AIC, BIC and growth rates for male broiler were 0.9998, 470.570, 21.681, 68.750, 68.934 and 0.241, respectively. The actual measured live weight values and the weight values estimated by Logistics, Gompertz, Weibull, Hossfeld, Von Bertalanffy models and MARS algorithm are close and in harmony with each other in the graph. However, the weight values estimated from the MARS algorithm are much closer to the observed live weight values. The represent study also demonstrated a very high predictive performance of the MARS data mining algorithm for describing the growth of chicken. In conclusion, MARS algorithm can be a good alternative for breeders aiming at describing the weight-age relationship of broiler chickens.

Most model selection procedures use a fixed penalty penalizing an increase in the size of a model. These nonadaptive selection procedures perform well only in one type of situation. For instance, Bayesian information criterion (BIC) with a large penalty performs well for “small” models and poorly for “large” models, and Akaike's information criterion (AIC) does just the opposite. This article proposes an adaptive model selection procedure that uses a data-adaptive complexity penalty based on a concept of generalized degrees of freedom. The proposed procedure, combining the benefit of a class of nonadaptive procedures, approximates the best performance of this class of procedures across a variety of different situations. This class includes many well-known procedures, such as AIC, BIC, Mallows's Cp, and risk inflation criterion (RIC). The proposed procedure is applied to wavelet thresholding in nonparametric regression and variable selection in least squares regression. Simulation results and an asymptotic analysis support the effectiveness of the proposed procedure.

EG goats are the result of crossing from male Ettawa goats and female Kacang goats so that these goats have properties between the two pedigrees (Setiaji et al., 2013).This goat has been raised by farmers widely in many regions in Indonesia due to good adaptability to the tropical environment, relatively rapid growth, and be utilised as meat and milk producer.Therefore, it is necessary to weigh

The age and growth information of fish is important for assessment of the dynamics, planning, and management in fisheries. In this study, the mean size at age data of 10 fish species collected from literature were fitted by the five growth models and the growth parameters were estimated by the maximum likelihood method. The best model was selected using the Akaike information criterion (AIC) and Bayesian information criterion (BIC). The growth of most species can be best described by the von Bertalanffy and allometric growth models. The results showed that both AIC and BIC have their advantages in the testing of significance of the difference between the functions of models. As for the demersal fishes (snappers and groupers), BIC is better than AIC in selecting the best growth model.

This article discusses the use of association models to detect group differences and the choice of model selection criteria under different conditions. The performances of several commonly used model selection criteria in log-linear modeling are examined using Monte Carlo simulations. The results suggest that no single criterion can play the role of panacea in model selection. L2/ df, normed fit index (NFI) (or 1 - L12/L02), and Akaike's information criterion (AIC) are systematically biased toward models incorporating group differences. The log-likelihood ratio test (LRT), in conjunction with the nested chi-square difference test, is useful for samples with moderate sizes. Bayesian information criterion (BIC) is most reliable among the measures tested. It consistently provides correct information about group differences in association, especially when the sample size is large. Although BIC may give equivocal results under certain conditions, the ambiguities can be mitigated by an inclusion of competing models and a cautious interpretation of any small improvement in BIC.

Latent profile analysis with nonnormal mixtures: A Monte Carlo examination of model selection using fit indices

Tong (1975) has proposed a procedure for estimating the order of a Markov chain based on Akaike's information criterion (AIC). In this paper, the asymptotic distribution of the AIC estimator is derived and it is shown that the estimator is inconsistent. As an alternative to the AIC procedure, the Bayesian information criterion (BIC) proposed by Schwarz (1978) is shown to be consistent. These two procedures yield different estimated orders when applied to specific samples of meteorological observations. For parameters based on these meteorological examples, the AIC and BIC procedures are compared by means of simulation for finite samples. The results obtained have practical implications concerning whether, in the routine fitting of precipitation data, it is necessary to consider higher than first-order Markov chains.

This study compared the performance of Poisson regression, negative binomial regression and generalized negative binomial regression models using road traffic crashes data collected from Ekiti state command of Federal Road Safety Commission (FRSC). The method of maximum likelihood estimation was employed to determine the parameter estimate for each models, the significance of each variable consider were examined. The criteria used in selecting the best model fitted are Akaike information criterion (AIC), Bayesian information criterion and Deviance. R software package was used in analysing the data. Firstly multicollinearity was tested to determine if there is high correlation between the predictor variables. Data were analysed and the result from the three model were compared using Akaike information criterion (AIC) and Bayesian information criterion (BIC) and the deviance.The result from the three models were compared using AIC, BIC and Deviance, AIC values for the models are (414.79, 476.8 and 587.312), BIC values (490.8873, 495.59 and 589.321) and Deviance value (61.93, 66.927 and 69.927) respectively. Generalized negative binomial regression (GNBR) had the least value of AIC and BIC, thus consider as the best model for analyzing road traffic crashes data in Ekiti state, Nigeria. Having shown a smaller AIC, BIC and deviance values, the generalized negative binomial was consider a better fit model when analysing road traffic crashes in Ekiti state, Nigeria.

We present a new Stata program, vselect, that helps users perform variable selection after performing a linear regression. Options for stepwise methods such as forward selection and backward elimination are provided. The user may specify Mallows's Cp, Akaike's information criterion, Akaike's corrected information criterion, Bayesian information criterion, or R2 adjusted as the information criterion for the selection. When the user specifies the best subset option, the leaps-and-bounds algorithm (Furnival and Wilson, Technometrics 16: 499–511) is used to determine the best subsets of each predictor size. All the previously mentioned information criteria are reported for each of these subsets. We also provide options for doing variable selection only on certain predictors (as in [R] nestreg) and support for weighted linear regression. All options are demonstrated on real datasets with varying numbers of predictors.