18 resultados para Variance.
em Deakin Research Online - Australia
Resumo:
This study investigates the determinants of the fertility rate in Taiwan over the period 1966–2001. Consistent with theory, the key explanatory variables in Taiwan's fertility model are real income, infant mortality rate, female education and female labor force participation rate. The test for cointegration is based on the recently developed bounds testing procedure while the long-run and short-run elasticities are based on the autoregressive distributed lag model. Among our key results, female education and female labor force participation rate are found to be the key determinants of fertility in Taiwan in the long run. The variance decom-position analysis indicates that in the long run approximately 45percent of the variation in fertility is explained by the combined impact of female labor force participation, mortality and income, implying that socioeconomic development played an important role in the fertility transition in Taiwan. This result is consistent with the traditional structural hypothesis.
Resumo:
One of the fundamental machine learning tasks is that of predictive classification. Given that organisations collect an ever increasing amount of data, predictive classification methods must be able to effectively and efficiently handle large amounts of data. However, it is understood that present requirements push existing algorithms to, and sometimes beyond, their limits since many classification prediction algorithms were designed when currently common data set sizes were beyond imagination. This has led to a significant amount of research into ways of making classification learning algorithms more effective and efficient. Although substantial progress has been made, a number of key questions have not been answered. This dissertation investigates two of these key questions. The first is whether different types of algorithms to those currently employed are required when using large data sets. This is answered by analysis of the way in which the bias plus variance decomposition of predictive classification error changes as training set size is increased. Experiments find that larger training sets require different types of algorithms to those currently used. Some insight into the characteristics of suitable algorithms is provided, and this may provide some direction for the development of future classification prediction algorithms which are specifically designed for use with large data sets. The second question investigated is that of the role of sampling in machine learning with large data sets. Sampling has long been used as a means of avoiding the need to scale up algorithms to suit the size of the data set by scaling down the size of the data sets to suit the algorithm. However, the costs of performing sampling have not been widely explored. Two popular sampling methods are compared with learning from all available data in terms of predictive accuracy, model complexity, and execution time. The comparison shows that sub-sampling generally products models with accuracy close to, and sometimes greater than, that obtainable from learning with all available data. This result suggests that it may be possible to develop algorithms that take advantage of the sub-sampling methodology to reduce the time required to infer a model while sacrificing little if any accuracy. Methods of improving effective and efficient learning via sampling are also investigated, and now sampling methodologies proposed. These methodologies include using a varying-proportion of instances to determine the next inference step and using a statistical calculation at each inference step to determine sufficient sample size. Experiments show that using a statistical calculation of sample size can not only substantially reduce execution time but can do so with only a small loss, and occasional gain, in accuracy. One of the common uses of sampling is in the construction of learning curves. Learning curves are often used to attempt to determine the optimal training size which will maximally reduce execution time while nut being detrimental to accuracy. An analysis of the performance of methods for detection of convergence of learning curves is performed, with the focus of the analysis on methods that calculate the gradient, of the tangent to the curve. Given that such methods can be susceptible to local accuracy plateaus, an investigation into the frequency of local plateaus is also performed. It is shown that local accuracy plateaus are a common occurrence, and that ensuring a small loss of accuracy often results in greater computational cost than learning from all available data. These results cast doubt over the applicability of gradient of tangent methods for detecting convergence, and of the viability of learning curves for reducing execution time in general.
Resumo:
Background :
The correlations between systolic blood pressure (SBP) and total cholesterol levels (CHOL) might result from genetic or environmental factors that determine variation in the phenotypes and are shared by family members. Based on 330 nuclear families in the Framingham Heart Study, we used a multivariate normal model, implemented in the software FISHER, to estimate genetic and shared environmental components of variation and genetic and shared environmental correlation between the phenotypes. The natural logarithm of the phenotypes measured at the last visit in both Cohort 1 and 2 was used in the analyses. The antihypertensive treatment effect was corrected before adjustment of the systolic blood pressure for age, sex, and cohort.
Results :
The univariate correlation coefficient was statistically significant for sibling pairs and parent-offspring pairs, but not significant for spouse pairs. In the bivariate analysis, the cross-trait correlation coefficients were not statistically significant for all relative pairs. The shared environmental correlation was statistically significant, but the genetic correlation was not significant.
Conclusion :
There is no significant evidence for a close genetic correlation between systolic blood pressure and total cholesterol levels. However, some shared environmental factors may determine the variation of both phenotypes.
Resumo:
The aim of the present study was to examine quantitative differences in lobar cerebral cortical volumes in a healthy adult population. Quantitative volumetric MRI of whole brain, cerebral and cerebellar volumes was performed in a cross-sectional analysis of 97 normal volunteers, with segmented frontal, temporal, parietal and occipital cortical volumes measured in a subgroup of 60 subjects, 30 male and 30 female, matched for age and sex. The right cerebral hemisphere was larger than the left across the study group with a small (<1%) but significant difference in symmetry (P < 0.001). No difference was found between volumes of right and left cerebellar hemispheres. Rightward cerebral cortical asymmetry (right larger than left) was found to be significant across all lobes except parietal. Males had greater cerebral, cerebellar and cerebral cortical lobar volumes than females. Larger male cerebral cortical volumes were seen in all lobes except for left parietal. Females had greater left parietal to left cerebral hemisphere and smaller left temporal to left cerebral hemisphere ratios. There was a mild reduction in cerebral volumes with age, more marked in males. This study confirms and augments past work indicating underlying structural asymmetries in the human brain, and provides further evidence that brain structures in humans are differentially sensitive to the effects of both age and sex.
Resumo:
Purpose - The purpose of this paper is to analyse the interdependencies of the house price growth rates in Australian capital cities.
Design/methodology/approach - A vector autoregression model and variance decomposition are introduced to estimate and interpret the interdependences among the growth rates of regional house prices in Australia.
Findings - The results suggest the eight capital cities can be divided into three groups: Sydney and Melbourne; Canberra, Adelaide and Brisbane; and Hobart, Perth and Darwin.
Originality/value - Based on the structural vector autoregression model, this research develops an innovative interdependence analysis approach of regional house prices based on a variance decomposition method.
Resumo:
Despite concern about method variance between measures as a bias in survey research, scholars have overlooked or ignored the effects of method variance within measures (i.e., covariation among items from the same scale that may be attributed to the method of measurement employed). Not only do few commonly used survey instruments reflect efforts to control for method variance, but guides to scale construction encourage researchers to implement strategies that enhance the effects of method variance within measures. In this article, we have argued that when method variance inflates relationships between questionnaire items, traditional psychometric indices overestimate the amount of true or construct variance that scales capture. Implications for survey research that uses fixed alternative questionnaire measures are delineated.
Resumo:
This paper proposes an innovative optimized parametric method for construction of prediction intervals (PIs) for uncertainty quantification. The mean-variance estimation (MVE) method employs two separate neural network (NN) models to estimate the mean and variance of targets. A new training method is developed in this study that adjusts parameters of NN models through minimization of a PI-based cost functions. A simulated annealing method is applied for minimization of the nonlinear non-differentiable cost function. The performance of the proposed method for PI construction is examined using monthly data sets taken from a wind farm in Australia. PIs for the wind farm power generation are constructed with five confidence levels between 50% and 90%. Demonstrated results indicate that valid PIs constructed using the optimized MVE method have a quality much better than the traditional MVE-based PIs.
Resumo:
Social foragers can alternate between searching for food (producer tactic), and searching for other individuals that have located food in order to join them (scrounger tactic). Both tactics yield equal rewards on average, but the rewards generated by producer are more variable. A dynamic variance-sensitive foraging model predicts that social foragers should increase their use of scrounger with increasing energy requirements and/or decreased food availability early in the foraging period. We tested whether natural variation in minimum energy requirements (basal metabolic rate or BMR) is associated with differences in the use of producer–scrounger foraging tactics in female zebra finches Taeniopygia guttata. As predicted by the dynamic variance-sensitive model, high BMR individuals had significantly greater use of the scrounger tactic compared with low BMR individuals. However, we observed no effect of food availability on tactic use, indicating that female zebra finches were not variance-sensitive foragers under our experimental conditions. This study is the first to report that variation in BMR within a species is associated with differences in foraging behaviour. BMR-related differences in scrounger tactic use are consistent with phenotype-dependent tactic use decisions. We suggest that BMR is correlated with another phenotypic trait which itself influences tactic use decisions.
Resumo:
Neural network (NN) is a popular artificial intelligence technique for solving complicated problems due to their inherent capabilities. However generalization in NN can be harmed by a number of factors including parameter's initialization, inappropriate network topology and setting parameters of the training process itself. Forecast combinations of NN models have the potential for improved generalization and lower training time. A weighted averaging based on Variance-Covariance method that assigns greater weight to the forecasts producing lower error, instead of equal weights is practiced in this paper. While implementing the method, combination of forecasts is done with all candidate models in one experiment and with the best selected models in another experiment. It is observed during the empirical analysis that forecasting accuracy is improved by combining the best individual NN models. Another finding of this study is that reducing the number of NN models increases the diversity and, hence, accuracy.
Resumo:
Continued population growth in Melbourne over the past decade has led to the development of a range of strategies and policies by State and Local levels of government to set an agenda for a more sustainable form of urban development. As the Victorian State government moves towards the development of 'Plan Melbourne', a new metropolitan planning strategy currently being prepared to take Melbourne forward to 2050, the following paper addresses the issue of how new residential built form will impact on and be accommodated in existing Inner Melbourne activity centres. Working with the prospect of establishing a more compact city in order to meet an inner city target of 90,000 new dwellings (Inner Metropolitan Action Plan - IMAP Strategy 5), the paper presents a 'Housing Variance Model' based on household structure and dwelling type. As capacity is progressively altered through a range of built form permutations, the research attempts to assess the impact on the urban morphology of a case study of four Major Activity Centres in the municipality of Port Phillip.
Resumo:
A statistical optimized technique for rapid development of reliable prediction intervals (PIs) is presented in this study. The mean-variance estimation (MVE) technique is employed here for quantification of uncertainties related with wind power predictions. In this method, two separate neural network models are used for estimation of wind power generation and its variance. A novel PI-based training algorithm is also presented to enhance the performance of the MVE method and improve the quality of PIs. For an in-depth analysis, comprehensive experiments are conducted with seasonal datasets taken from three geographically dispersed wind farms in Australia. Five confidence levels of PIs are between 50% and 90%. Obtained results show while both traditional and optimized PIs are hypothetically valid, the optimized PIs are much more informative than the traditional MVE PIs. The informativeness of these PIs paves the way for their application in trouble-free operation and smooth integration of wind farms into energy systems. © 2014 Elsevier Ltd. All rights reserved.