Information Criterion For Selection Research Articles

Post-weaning survival in kangaroos is high and constant until senescence: Implications for population dynamics.

Large herbivores typically have consistently high prime-aged adult survival and lower, more variable, juvenile, and senescent survival. Many kangaroo populations undergo greater fluctuations in density compared with other large herbivores, but age- and sex-specific survival of kangaroos and their response to environmental variation remain poorly estimated. We used long-term capture-mark-recapture data on 920 individuals to investigate the survival component of eastern grey kangaroo (Macropus giganteus) population dynamics. Forage availability and population density were monitored quarterly and included as predictors of survival in Bayesian Cormack-Jolly-Seber models. Annual survival probabilities were estimated for five age classes: 0 years (juveniles), 1-2 years (subadults), 3-6 years (prime-aged adults), 7-9 years (presenescent adults), and ≥10 years (senescent adults). Survival of juveniles varied widely during our 12-year study, ranging from 0.07 to 0.90 for females and 0.05-0.92 for males. Subadult survival was 0.80-0.93 for females and 0.75-0.85 for males, while that of prime-aged adults was ≥0.94 for females and ≥0.83 for males, despite large fluctuations in forage and density. The survival of presenescent adults spanned 0.86-0.93 for females and 0.60-0.86 for males. Senescent survival was variable, at 0.49-0.90 for females and 0.49-0.80 for males. Male survival was significantly lower than female survival in prime-aged and presenescent adults, but not in other age classes. Although most of the models supported by Watanabe-Akaike Information Criterion selection included at least one environmental covariate, none of these covariates individually had a discernible effect on survival. Temporal variability in overall survival appeared mostly due to changes in the survival of juvenile and senescent kangaroos. Kangaroo survival patterns are similar to those of ungulates, suggesting a strong role of sex-age structure on population dynamics.

Evaluating Different Selection Criteria for Phase Type Survival Tree Construction

Due to its interpretability and intuitiveness, survival tree based analysis is a powerful Artificial Intelligence method for modelling longitudinal survival data, its relationship with covariates and the interrelationship between covariates. Furthermore, it is being increasingly used for a range of applications including clustering, prognostication and classification. Phase type survival tree methods have been demonstrated to have important applications, including clustering patients into clinically meaningful groups, patient pathway prognostication and forecasting bed requirements. In this article, we critically investigate and assess several selection information criteria with regards to their suitability and limitations when used as splitting criteria in phase type survival tree construction. As shown in Table 12, the results of this analysis are compared and discussed. Furthermore, a text mining approach is utilised to further assess correlations, which have been extracted from hospital data, between the three underlying diseases and the two different types of population groups, namely age and gender groups. Its aim is to provide further investigative tools. In fact, due to its ability to analyse large volumes of textual data, text mining can provide a useful approach to this research area.

An Exponentiated Odd Lindley Inverse Exponential Distribution and its Application to Infant Mortality and HIV Transmission Rates in Nigeria

Recently, researchers have shown much interest in developing new continuous probability distributions by adding one or two parameter(s) to the some existing baseline distributions. This act has been beneficial to the field of statistical theory especially in modeling of real life situations. Also, the exponentiated family as used in developing new distributions is an efficient method proposed and studied for defining more flexible continuous probability distributions for modeling real life data. In this study, the method of exponentiation has been used to develop a new distribution called “Exponentiated odd Lindley inverse exponential distribution”. Some properties of the proposed distribution and estimation of its unknown parameters has been done using the method of maximum likelihood estimation and its application to real life datasets. The new model has been applied to infant mortality rate and mother-to-child HIV transmission rate. The results of these two applications reveal that the proposed model is a better model compared to the other fitted existing models by some selection information criteria.

Alanine Aminotransferase and Gamma-Glutamyl Transpeptidase Predict Histologic Improvement in Pediatric Nonalcoholic Steatohepatitis.

Predictive, noninvasive tools are needed to monitor key features of nonalcoholic fatty liver disease (NAFLD) in children that relate to improvement in liver histology. The purpose of this study was to evaluate the relationship between liver chemistries and liver histology using data from the CyNCh (Cysteamine Bitartrate Delayed-Release for the Treatment of NAFLD in Children) clinical trial. This study included 146 children. Improvement in liver histology, defined as decrease in nonalcoholic fatty liver disease (NAFLD) Activity Score ≥2 points without worsening of fibrosis, occurred in 43 participants (30%). There were 46 participants with borderline zone 1 nonalcoholic steatohepatitis (NASH) at baseline, with resolution in 28% (12 of 46). Multivariate models were constructed using baseline and change in alanine aminotransferase (ALT), aspartate aminotransferase (AST), and gamma-glutamyl transferase (GGT) at 52weeks, for improvement in (1) liver histology primary outcome, (2) borderline zone 1 NASH, and (3) fibrosis. For improvement in histology, the model (P<0.0001) retained baseline and change in GGT (area under the receiver operating characteristic [AUROC], 0.79; 95% confidence interval [CI], 0.71-0.87). For borderline zone 1 NASH, the model (P=0.0004) retained baseline and change in ALT (AUROC, 0.80; 95% CI, 0.67-0.93). For fibrosis, the model (P<0.001) retained baseline and change in ALT (AUROC, 0.80; 95% CI, 0.67-0.93). Additional clinical parameters were added to the models using Akaike's information criterion selection, and significantly boosted performance: improvement in histology with AUROC of 0.89 (95% CI, 0.82-0.95), borderline zone 1 NASH with AUROC of 0.91 (95% CI, 0.83-0.99), and fibrosis with AUROC of 0.89 (95% CI, 0.82-0.94). Models were validated using data from the TONIC (Treatment of Nonalcoholic Fatty Liver Disease in Children) trial. In children with NAFLD, dynamic changes in serum ALT and GGT are associated with change in liver histology and appear to be powerful indicators of histological response.

Modeling Two Dimensional Touch Pointing.

Modeling touch pointing is essential to touchscreen interface development and research, as pointing is one of the most basic and common touch actions users perform on touchscreen devices. Finger-Fitts Law [4] revised the conventional Fitts' law into a 1D (one-dimensional) pointing model for finger touch by explicitly accounting for the fat finger ambiguity (absolute error) problem which was unaccounted for in the original Fitts' law. We generalize Finger-Fitts law to 2D touch pointing by solving two critical problems. First, we extend two of the most successful 2D Fitts law forms to accommodate finger ambiguity. Second, we discovered that using nominal target width and height is a conceptually simple yet effective approach for defining amplitude and directional constraints for 2D touch pointing across different movement directions. The evaluation shows our derived 2D Finger-Fitts law models can be both principled and powerful. Specifically, they outperformed the existing 2D Fitts' laws, as measured by the regression coefficient and model selection information criteria (e.g., Akaike Information Criterion) considering the number of parameters. Finally, 2D Finger-Fitts laws also advance our understanding of touch pointing and thereby serve as the basis for touch interface designs.

Online updating of information based model selection in the big data setting

The generalized information criterion (GIC) is an important tool for model selection in statistical inference. In the big data setting, traditional GIC cannot be calculated when the data size exceeds the computer memory. We propose an online updating approach to calculate the GIC, and perform model selection for huge datasets. Specifically, we define the online updating versions of GICs for streaming data for the normal linear regression and generalized linear models. Under reasonable regularity conditions, we show that the information criterion selection procedures are asymptotically valid. The performance of the proposed criteria is assessed using extensive simulation study. The usage of our proposed model selection procedure is further illustrated with the analysis of two large datasets, the covertype data and the earthquake data. For both datasets, the online updating procedure selected the same or similar model as the entire data based model selection procedure.

Empirical-likelihood-based criteria for model selection on marginal analysis of longitudinal data with dropout missingness.

Longitudinal data are common in clinical trials and observational studies, where missing outcomes due to dropouts are always encountered. Under such context with the assumption of missing at random, the weighted generalized estimating equation (WGEE) approach is widely adopted for marginal analysis. Model selection on marginal mean regression is a crucial aspect of data analysis, and identifying an appropriate correlation structure for model fitting may also be of interest and importance. However, the existing information criteria for model selection in WGEE have limitations, such as separate criteria for the selection of marginal mean and correlation structures, unsatisfactory selection performance in small-sample setups, and so forth. In particular, there are few studies to develop joint information criteria for selection of both marginal mean and correlation structures. In this work, by embedding empirical likelihood into the WGEE framework, we propose two innovative information criteria named a joint empirical Akaike information criterion and a joint empirical Bayesian information criterion, which can simultaneously select the variables for marginal mean regression and also correlation structure. Through extensive simulation studies, these empirical-likelihood-based criteria exhibit robustness, flexibility, and outperformance compared to the other criteria including the weighted quasi-likelihood under the independence model criterion, the missing longitudinal information criterion, and the joint longitudinal information criterion. In addition, we provide a theoretical justification of our proposed criteria, and present two real data examples in practice for further illustration.

The Impact of Transportation Infrastructure on Economic Growth: Empirical Evidence from Saudi Arabia

The transport sector is an important factor of economic activity, where it contributes directly to economic activities and employment. The road has a large indirect impact on all the other sectors and activities in the economy. The study aims to investigate the causality relations between road land and economic growth in Saudi Arabia. The study is based on secondary data gained from Saudi Arabia Monetary Agency and World Bank over the period of1988 to2017. The Granger causality test was used to investigate the relationship between the variables with Akiake Lag Length Selection Information Criteria, while Vector Autoregtression (VAR) model was used in order to find the causality. The result reveals unidirectional causality form real GDP to road; however, there is no evidence to support that transportation infrastructure is the cause of economic growth. Granger causality from GDP to investment in infrastructure indicates that reinvestment in infrastructure is caused by economic growth and not vice versa. Economic growth drove pressures on existing transport infrastructure and required additional investment. The finding is in line with the commonly accepted notion advocating that economic growth or development provide necessary financial and technical support for transportation infrastructure investment and improvement.

Weighted Average Information Criterion for Selection of an Asymmetric Price Relationship

TThis study evaluates the performance of the recently developed model selection criteria (WIC) against commonly used alternatives (AIC and BIC) in terms of their ability to recover the true asymmetric data generating process. Monte Carlo simulation results indicate that the performance of the model selection methods depends on the sample size, the difference in asymmetric adjustment parameters and the amount of noise in the model used in the application. WIC outperforms AIC and BIC under stable conditions such as a large sample and small noise levels. Additionally, WIC outperforms AIC and BIC as the difference between asymmetric adjustment speeds increases. These results suggest that WIC is a very reliable and useful criterion in asymmetric price transmission model selection.

Characterizing larval swordfish habitat in the western tropical North Atlantic

AbstractSwordfish Xiphias gladius (Linnaeus, 1758) are a circumglobal pelagic fish targeted by multiple lucrative fisheries. Determining the distribution of swordfish larvae is important for indicating reproductive activity and understanding the early life history of swordfish. We identify and characterize larval swordfish distributions during peak swordfish spawning throughout the Gulf of Mexico and western Caribbean Sea with generalized additive models (GAMs) using catches of swordfish larvae during ichthyoplankton surveys in April and May of 2010, 2011, and 2012. The best fit GAM, as determined by stepwise, backward Akaike Information Criterion selection, included both physiochemical (temperature at 5 m, sea surface height anomaly (SSHA), eddy kinetic energy (EKE)), temporal (lunar illumination, hour of sampling) and spatial (location) variables, while near surface chlorophyll a concentration residuals remained as a random effect. The highest probability of larval swordfish catch occurred at sub‐surface temperatures, SSHA, and EKE values indicative of boundary currents. Standard lengths of larvae were larger further downstream in the boundary currents, despite high variability in length with location due to multiple spawning locations of swordfish near these currents. Probability of larval swordfish catch also peaked during the crescent and gibbous moons, indicating a lunar periodicity to swordfish spawning. These results suggest that swordfish may spawn during select moon phases near boundary currents that transport their larvae to larval and juvenile habitat including the northern Gulf of Mexico and coastal waters of the southeast United States.

Performance of information criteria for selection of Hawkes process models of financial data

We test three common information criteria (IC) for selecting the order of a Hawkes process with an intensity kernel that can be expressed as a mixture of exponential terms. These processes find application in high-frequency financial data modelling. The information criteria are Akaike’s information criterion, the Bayesian information criterion and the Hannan–Quinn criterion. Since we work with simulated data, we are able to measure the performance of model selection by the success rate of the IC in selecting the model that was used to generate the data. In particular, we are interested in the relation between correct model selection and underlying sample size. The analysis includes realistic sample sizes and parameter sets from recent literature where parameters were estimated using empirical financial intra-day data. We compare our results to theoretical predictions and similar empirical findings on the asymptotic distribution of model selection for consistent and inconsistent IC.

Performance of Information Criteria for Selection of Hawkes Process Models of Financial Data

We test three common information criteria (IC) for selecting the order of a Hawkes process with an intensity kernel that can be expressed as a mixture of exponential terms. These processes find application in high-frequency financial data modelling. The information criteria are Akaike’s information criterion (AIC), the Bayesian information criterion (BIC) and the Hannan-Quinn criterion (HQ). Since we work with simulated data, we are able to measure the performance of model selection by the success rate of the IC in selecting the model that was used to generate the data. In particular, we are interested in the relation between correct model selection and underlying sample size. The analysis includes realistic sample sizes and parameter sets from recent literature where parameters were estimated using empirical financial intra-day data. We compare our results to theoretical predictions and similar empirical findings on the asymptotic distribution of model selection for consistent and inconsistent IC.

Impact of Laws Regulating Pain Clinics on Opioid Prescribing and Opioid-Related Toxicity Among Texas Medicare Part D Beneficiaries.

Pain management clinics are major sources of prescription opioids. Texas government passed several laws regulating pain clinics between 2009 and 2011 to reduce opioid-related toxicity. Understanding the impact of these laws can inform policy geared toward making the laws more effective in curbing the growing epidemic of opioid overdose, especially among the elderly population. To examine the longitudinal association of laws regulating pain clinics on opioid-prescribing and opioid-related toxicity among Texas Medicare recipients. The 2007 to 2012 claims data for Texas Medicare Part D recipients were used to assess temporal trends in the percentage of patients filling any schedule II or schedule III opioid prescription, hospitalization for opioid toxicity, and their relationships to the 2009 to 2011 Texas laws regulating pain clinics. We excluded those with a cancer diagnosis. Join-point trend analysis with Bayesian Information Criterion selection methods were used to evaluate the change in monthly percentages of patients filling opioid prescriptions and hospitalization over time. There was a short-lived decline in the monthly percentages of patients who filled a schedule II or schedule III opioid prescription after the 2009 laws regulating pain clinics. The decline lasted about 3 months. Subsequent new laws had no effect on the percentages of patients who filled any opioid prescription or were hospitalized for potential opioid toxicity. Hospitalizations for opioid toxicity were highest in the winter and lowest in the summer. Changes in the percentages of opioid-prescribing or opioid-related hospitalizations over time were not associated with laws regulating pain clinics.

Estimation of the brown bear population on the Kenai Peninsula, Alaska

ABSTRACTThe brown bear population on the Kenai Peninsula, Alaska, has not been empirically estimated previously because conventional aerial methods over this heavily forested landscape were infeasible. We applied a rapid field protocol to a DNA‐based, mark‐recapture approach on a large and tightly bounded sample frame to estimate brown bear abundance. We used lure to attract bears to barbed wire stations deployed in 145 9‐km × 9‐km cells systematically distributed across 10,200 km2 of available habitat on the Kenai National Wildlife Refuge and Chugach National Forest during 31 consecutive days in early summer 2010. Using 2 helicopters and 4 2‐person field crews, we deployed the stations during a 6‐day period and subsequently revisited these stations on 5 consecutive 5‐day trap sessions. We extracted DNA to identify individual bears and developed capture histories as input to mark‐recapture models. Combined with data from radio‐telemetered bears, ≥243 brown bears were alive on the Kenai Peninsula in 2010, but we used only 99 females and 104 males in modeling. We used Akaike's Information Criterion selection and model averaging to estimate 428 (95% lognormal CI = 353–539) brown bears (including all age classes) on the study area. Despite low recaptures rates, we achieved reasonable precision by ensuring geographic and demographic population closure through a spatially comprehensive sample frame and very short sampling window. We reduced bias by including information from rub trees and telemetered females (i.e., occasion 0). Extrapolating the density estimate of 42 bears/1,000 km2 of available habitat on the study area to the Kenai Peninsula suggests a peninsula‐wide population of 582 brown bears (95% lognormal CI = 469−719). Despite a density estimate that is low compared to other coastal brown bear populations in Alaska and genetic evidence that suggests this peninsular population is insular, harvest management has been liberalized since 2012. We recommend this population estimate serve as the benchmark for future management. Published 2015. This article is a U.S. Government work and is in the public domain in the USA.

Model selection in the presence of incidental parameters

This paper considers model selection in panels where incidental parameters are present. Primary interest centers on selecting a model that best approximates the underlying structure involving parameters that are common within the panel. It is well known that conventional model selection procedures are often inconsistent in panel models and this can be so even without nuisance parameters. Modifications are then needed to achieve consistency. New model selection information criteria are developed here that use either the Kullback–Leibler information criterion based on the profile likelihood or the Bayes factor based on the integrated likelihood with a bias-reducing prior. These model selection criteria impose heavier penalties than those associated with standard information criteria such as AIC and BIC. The additional penalty, which is data-dependent, properly reflects the model complexity arising from the presence of incidental parameters. A particular example is studied in detail involving lag order selection in dynamic panel models with fixed effects. The new criteria are shown to control for over/under-selection probabilities in these models and lead to consistent order selection criteria.

The perils of quasi‐likelihood information criteria

In this paper, we consider some potential pitfalls of the growing use of quasi‐likelihood‐based information criteria for longitudinal data to select a working correlation structure in a generalized estimating equation framework. In particular, we examine settings where the fully conditional mean does not equal the marginal mean as well as hypothesis testing following selection of the working correlation matrix. Our results suggest that the use of any information criterion for selection of the working correlation matrix is inappropriate when the conditional mean model assumption is violated. We also find that type I error differs from the nominal level in moderate sample sizes following selection of the form of the working correlation but improves as sample size is increased as the selection is then concentrated on a single correlation structure. Our results serve to underline the potential dangers that can arise when using information criteria to select correlation structure in routine data analysis. Copyright © 2015 John Wiley &amp; Sons, Ltd.

High-dimensional Ising model selection with Bayesian information criteria

We consider the use of Bayesian information criteria for selection of the graph underlying an Ising model. In an Ising model, the full conditional distributions of each variable form logistic regression models, and variable selection techniques for regression allow one to identify the neighborhood of each node and, thus, the entire graph. We prove high-dimensional consistency results for this pseudo-likelihood approach to graph selection when using Bayesian information criteria for the variable selection problems in the logistic regressions. The results pertain to scenarios of sparsity, and following related prior work the information criteria we consider incorporate an explicit prior that encourages sparsity.

Model Selection Information Criteria in Latent Class Models with Missing Data and Contingency Question

Latent class analysis (LCA) has been found to have important applications in social and behavioral sciences for modeling categorical response variables, and nonresponse is typical when collecting data. In this study, the nonresponse mainly included “contingency questions” and real “missing data.” The primary objective of this research was to evaluate the effects of some potential factors on model selection indices in LCA with nonresponse data.We simulated missing data with contingency questions and evaluated the accuracy rates of eight information criteria for selecting the correct models. The results showed that the main factors are latent class proportions, conditional probabilities, sample size, the number of items, the missing data rate, and the contingency data rate. Interactions of the conditional probabilities with class proportions, sample size, and the number of items are also significant. From our simulation results, the impact of missing data and contingency questions can be amended by increasing the sample size or the number of items.

A New Criterion for Lag-Length Selection in Unit Root Tests

This paper examines lag selection problem in unit root tests which has become a major specification problem in empirical analysis of non-stationary time series data. It is known that the implementation of unit root tests requires the choice of optimal truncation lag for good power proper ties and it is equally unrealistic to assume that the true optimal truncation lag is known a prior to the practitioners and other applied researchers. Consequently, these users rely largely on the use of standard information criteria for selection of truncation lag in unit root tests. A number of previous studies have shown that these criteria have problem of over-specification of truncation lag-length leading to the well-known low power problem that is commonly associated with most unit root tests in the literature. This paper focuses on the problem of over-specification of truncation lag-length within the context of augmented Dickey-Fuller (ADF) and generalized least squares Dickey-Fuller (DF-GLS)unit root tests. In an attempt to address this lag selection problem, we propose a new criterion for the selection of truncation lag in unit root tests based on Koyck distributed lag model and we show that this new criterion avoids the problem of over-specification of truncationlag-length that is commonly associated with standard information criteria.

Model Selection in the Presence of Incidental Parameters

This paper considers model selection in nonlinear panel data models where incidental parameters or large-dimensional nuisance parameters are present. Primary interest typically centres on selecting a model that best approximates the underlying structure involving parameters that are common within the panel after concentrating out the incidental parameters. It is well known that conventional model selection procedures are often inconsistent in panel models and this can be so even without nuisance parameters (Han et al, 2012). Modifications are then needed to achieve consistency. New model selection information criteria are developed here that use either the Kullback-Leibler information criterion based on the profile likelihood or the Bayes factor based on the integrated likelihood with the robust prior of Arellano and Bonhomme (2009). These model selection criteria impose heavier penalties than those associated with standard information criteria such as AIC and BIC. The additional penalty, which is data-dependent, properly reflects the model complexity arising from the presence of incidental parameters. A particular example is studied in detail involving lag order selection in dynamic panel models with fixed individual effects. The new criteria are shown to control for over/under-selection probabilities in these models and lead to consistent order selection criteria.