OWA operators in linear regression and detection of outliers

We consider the use of Ordered Weighted Averaging (OWA) in linear regression. Our goal is to replace the traditional least squares, least absolute deviation, and maximum likelihood criteria with an OWA function of the residuals. We obtain several high breakdown robust regression methods as special cases (least median, least trimmed squares, trimmed likelihood methods). We also present new formulations of regression problem. OWA-based regression is particularly useful in the presence of outliers.

OWA operators in regression problems

We consider an application of fuzzy logic connectives to statistical regression. We replace the standard least squares, least absolute deviation, and maximum likelihood criteria with an ordered weighted averaging (OWA) function of the residuals. Depending on the choice of the weights, we obtain the standard regression problems, high-breakdown robust methods (least median, least trimmed squares, and trimmed likelihood methods), as well as new formulations. We present various approaches to numerical solution of such regression problems. OWA-based regression is particularly useful in the presence of outliers, and we illustrate the performance of the new methods on several instances of linear regression problems with multiple outliers.

Maximum likelihood estimation for rician distributed data in analytical q-ball imaging

Analytical q-ball imaging is widely used for reconstruction of orientation distribution function (ODF) using diffusion weighted MRI data. Estimating the spherical harmonic coefficients is a critical step in this method. Least squares (LS) is widely used for this purpose assuming the noise to be additive Gaussian. However, Rician noise is considered as a more appropriate model to describe noise in MR signal. Therefore, the current estimation techniques are valid only for high SNRs with Gaussian distribution approximating the Rician distribution. The aim of this study is to present an estimation approach considering the actual distribution of the data to provide reliable results particularly for the case of low SNR values. Maximum likelihood (ML) is investigated as a more effective estimation method. However, no closed form estimator is presented as the estimator becomes nonlinear for the noise assumption of the Rician distribution. Consequently, the results of LS estimator is used as an initial guess and the more refined answer is achieved using iterative numerical methods. According to the results, the ODFs reconstructed from low SNR data are in close agreement with ODFs reconstructed from high SNRs when Rician distribution is considered. Also, the error between the estimated and actual fiber orientations was compared using ML and LS estimator. In low SNRs, ML estimator achieves less error compared to the LS estimator.

Management control systems use, organisational capabilities and performance in universities: application of the resource-based view

Invoking the resource-based view (RBV), this study investigates relationships between management control systems (MCSs) use, including information use from performance measurement systems (PMSs), and organisational capabilities in the context of academic units of Australian universities. Increased competition and attention to distinctive capabilities amongst universities, particularly at their strategic operating unit level of Schools 1, provides the setting for application of this theoretic perspective. The objective of this study is to model various relationships between diagnostic and interactive use of MCSs, attention given to centrally-imposed and discretionary types of PMS information, the strength of capabilities of the academic unit and, in turn, performance of the academic units. This objective is investigated using a field survey in which a mail survey instrument is administered to a census of all Heads of Schools in all 39 universities in Australia. Valid responses were received from 166 Heads. Principal components factor analysis finds that Heads conceived capabilities of their unit in functional dimensions, not in generic dimensions as found in prior literature; Heads also considered performance measures in terms of their importance (critical or discretionary) rather than type (financial versus non-financial). Partial least-squares analysis is then used for path modelling, and several significant results are obtained. Highlights are that diagnostic MCS use and centrally-imposed performance measures, i.e., key performance indicators, but not interactive MCS use or discretionary performance measures, significantly relate to some or all of the strength of capabilities in the fields of teaching, research and networking, and in turn indirectly relate to performance of the academic units. The findings have practical implications for styles of control systems use; focus on selected key performance measures; and development of organisational capabilities for achievement of superior performance by academic schools in universities.

The pricing of infrastructure initial public offerings : evidence from Australia

This paper explores first-day returns on infrastructure entity initial public offerings (IPOs) in Australia from 1996 to 2007. While a good deal has been written on the first-day returns of industrial and mining company IPOs and Real Estate Investment Trust IPOs, first-day returns of infrastructure entity IPOs have yet to be reported in the literature. The study uses ordinary least squares regression analysis to identify factors that might influence the percentage first-day returns theoretically available to investing subscribers and factors that might influence the aggregate amount of money left to subscribers by issuers. The study finds that first-day returns, on average, are not significantly different from zero. There is evidence, however, that suggests higher dividend yields and higher percentage direct costs of capital raising influence these first-day returns. The study also finds that infrastructure entity IPOs that seek to raise more equity capital leave less money on the table for subscribing investors.

N-acetyl cysteine for depressive symptoms in bipolar disorder - a double blind, randomized, placebo-controlled trial

Background: Treatment-resistant subthreshold depression is a major problem in bipolar disorder. Both depression and bipolar disorderare complicated by glutathione depletion. We hypothesized that treatment with N-acetyl cysteine (NAC), a safe, orally bioavailable precursor of glutathione, may improve the depressive component of bipolar disorder.

Methods: A randomized, double-blind, multicenter, placebo-controlled study of individuals (n 75) with bipolar disorder in the maintenance phase treated with NAC (1 g twice daily) adjunctive to usual medication over 24 weeks, with a 4-week washout. The two primary outcomes were the Montgomery Asberg Depression Rating Scale (MADRS) and time to a mood episode. Secondary outcomes included the Bipolar Depression Rating Scale and 11 other ratings of clinical status, quality of life, and functioning.

Results: NAC treatment caused a significant improvement on the MADRS (least squares mean difference [95% confidence interval]: 8.05 [13.16, 2.95], p .002) a n d most secondary scales at end point. Benefit was evident by 8 weeks on the Global Assessment of Functioning Scale and Social and Occupational Functioning Assessment Scale and at 20 weeks on the MADRS. Improvements were lost after washout. There was no effect of NAC on time to a mood episode (log-rank test: p .968) and no significant between-group differences inadverse events. Effect sizes at end point were medium to high for improvements in MADRS and 9 of the 12 secondary readouts.

Conclusions: NAC appears a safe and effective augmentation strategy for depressive symptoms in bipolar  disorder.

Agility in information systems : enabling capabilities for the IT function

This research identifies how the IT function can create agility in existing information systems. Agility is the capability to quickly sense and respond to environmental perturbations. We contrasted perspectives on agility from a widely used industry framework and that of the IS research literature. Beer’s Viable System Model was a useful meta-level theory to house agility elements from IS research and it introduced cybernetic principles to identify the processes required of the IT function. Indeed, our surveys of 70 organizations confirmed that the applied theory better correlates with reported agility than does existing industry best practice.

The research conducted two quantitative surveys to test the applied theory. The first survey mailed a Likert-type questionnaire to the clients of an Australian IT consultancy. The second survey invited international members of professional interest groups to complete a web-based questionnaire. The responses from the surveys were analyzed using partial-least-squares modeling. The data analysis positively correlated the maturity of IT function processes prescribed by the VSM and the likelihood of agility in existing information systems. We claim our findings generalize to other large organizations in OECD member countries.

The research offers an agility-capability model of the IT function to explain and predict agility in existing information systems. A further contribution is to improve industry ‘best practice’ frameworks by prescribing processes of the IT function to develop in maturity.

Immobilized artificial membrane chromatography : quantitative structure-retention relationships of structurally diverse drugs

The chromatographic capacity factors (log k‘) for 32 structurally diverse drugs were determined by high performance liquid chromatography (HPLC) on a stationary phase composed of phospholipids, the so-called immobilized artificial membrane (IAM). In addition, quantitative structure-retention relationships (QSRR) were developed in order to explain the dependence of retention on the chemical structure of the neutral, acidic, and basic drugs considered in this study. The obtained retention data were modeled by means of multiple regression analysis (MLR) and partial least squares (PLS) techniques. The structures of the compounds under study were characterized by means of calculated physicochemical properties and several nonempirical descriptors. For the carboxylic compounds included in the analysis, the obtained results suggest that the IAM-retention is governed by hydrophobicity factors followed by electronic effects due to polarizability in second place. Further, from the analysis of the results obtained of two developed quantitative structure-permeability studies for 20 miscellaneous carboxylic compounds, it may be concluded that the balance between polarizability and hydrophobic effects is not the same toward IAM phases and biological membranes. These results suggest that the IAM phases could not be a suitable model in assessing the acid-membrane interactions. However, it is not possible to generalize this observation, and further work in this area needs to be done to obtain a full understanding of the partitioning of carboxylic compounds in biological membranes. For the non-carboxylic compounds included in the analysis, this work shows that the hydrophobic factors are of prime importance for the IAM-retention of these compounds, while the specific polar interactions, such as electron pair donor−acceptor interactions and electrostatic interactions, are also involved, but they are not dominant.

Assessment of spatiotemporal varying relationships between rainfall, land cover and surface water area using geographically weighted regression

Traditional regression techniques such as ordinary least squares (OLS) are often unable to accurately model spatially varying data and may ignore or hide local variations in model coefficients. A relatively new technique, geographically weighted regression (GWR) has been shown to greatly improve model performance compared to OLS in terms of higher R 2 and lower corrected Akaike information criterion (AICC). GWR models have the potential to improve reliabilities of the identified relationships by reducing spatial autocorrelations and by accounting for local variations and spatial non-stationarity between dependent and independent variables. In this study, GWR was used to examine the relationship between land cover, rainfall and surface water habitat in 149 sub-catchments in a predominately agricultural region covering 2.6 million ha in southeast Australia. The application of the GWR models revealed that the relationships between land cover, rainfall and surface water habitat display significant spatial non-stationarity. GWR showed improvements over analogous OLS models in terms of higher R 2 and lower AICC. The increased explanatory power of GWR was confirmed by the results of an approximate likelihood ratio test, which showed statistically significant improvements over analogous OLS models. The models suggest that the amount of surface water area in the landscape is related to anthropogenic drainage practices enhancing runoff to facilitate intensive agriculture and increased plantation forestry. However, with some key variables not present in our analysis, the strength of this relationship could not be qualified. GWR techniques have the potential to serve as a useful tool for environmental research and management across a broad range of scales for the investigation of spatially varying relationships.

High pressure measurements of the refractive indices of two nematic liquid crystals

The refractive indices of two nematogens, 4-methoxy-benzylidene-4 prime -n-butylaniline (MBBA) and 4-n-pentyl-4 prime -cyanobiphenyl (5CB), were measured throughout their nematic ranges at pressures up to 2 kbar and temperatures up to 70 degree C in the first substance and up to 5 kbar and 145 degree C in the second. Measurements were made at lambda equals 5,890 A, using a sensitive interference fringe technique. Results are presented in the form of functions n//e(P, T) for the extraordinary index and n//o (P, T) for the ordinary index, obtained by least squares fits to the experimental data.

The level of representative underwriting and the underwriting costs of REIT SEOs

Purpose – The purpose of this paper is to investigate factors influencing the underwriting discount for US Real Estate Investment Trust (REIT) Seasoned Equity Offerings (SEOs).

Design/methodology/approach – The study provides new evidence on determinants of underwriting discounts with a comprehensive dataset of 783 US REIT SEOs from 1996 until June 2010. Ordinary least squares regressions are performed to estimate the effect of the level of representative underwriting along with other potential factors on underwriting discounts.

Findings – The study complements the well-documented notion of the economies of scale in SEO underwriting discounts. The equally (value) weighted underwriting discounts averaged 4.21 per cent (4.10 per cent) with a declining trend over time. The findings of this study show the statistically and economically significant negative effect of the level of representative underwriting on the underwriting discounts, as well as the significance of the structure of underwriting syndicate in determining the underwriting discounts. The findings suggest that issuers can minimize the costs of raising secondary equity capital by optimally allocating the underwriting business among the underwriters.

Originality/value – This paper adds to the international REIT SEO literature by exploring new evidence behind underwriting discounts. The study includes data before and after the REIT Modernization Act 1999 and during the recent global financial crisis period.

Reproductive consequences of male arrival order in the Bark beetle, Ips grandicollis

The ability of birds to perceive, assess and appropriately respond to the presence of relatively novel threats is important to their survival. We hypothesized that the cognitive capacity of birds will influence their ability for accurate response to novelty. We used brain volume as a surrogate for cognitive capacity and postulated that larger brained birds would moderate their responses when presented with a benign, frequently occurring stimulus, such as a person, because they would habituate more readily. We conducted phylogenetic generalized least square regression to investigate the relationship between brain volume and flight initiation distance (FID; the distance to which a bird can be approached before initiating escape behaviour), while controlling for confounding factors including body size (body mass and wing length) and migration status. We compared seven different models using combinations of these parameters using Akaike's information criterion to determine the best approximating model(s) explaining FID. The two best-supported models included only wing length and only body mass with Akaike weights of 0.396 and 0.311 respectively. No model including brain volume had an Akaike weight greater than 0.083 and brain volume was poorly correlated with FID in models after controlling for body mass. Thus, brain volume does not appear to strongly relate to bravery among these shorebirds.

Modeling of a stereo vision system using a genetic algorithm based fuzzy linear regression

In this paper a fuzzy linear regression (FLR) model integrated with a genetic algorithm (GA) is proposed. The proposed GA-FLR model is applied to modeling of a stereo vision system. A set of empirical data from stereo vision object measurement is collected based on the full factorial design technique. Three regression models, namely ordinary least-squares regression (OLS), FLR, and GA-FLR, are developed, and with their performances compared. The results show that the proposed GA-FLR model performs better than OLS and FLR in modeling of a stereo vision system.

Derivative-free optimization and neural networks for robust regression

Large outliers break down linear and nonlinear regression models. Robust regression methods allow one to filter out the outliers when building a model. By replacing the traditional least squares criterion with the least trimmed squares (LTS) criterion, in which half of data is treated as potential outliers, one can fit accurate regression models to strongly contaminated data. High-breakdown methods have become very well established in linear regression, but have started being applied for non-linear regression only recently. In this work, we examine the problem of fitting artificial neural networks (ANNs) to contaminated data using LTS criterion. We introduce a penalized LTS criterion which prevents unnecessary removal of valid data. Training of ANNs leads to a challenging non-smooth global optimization problem. We compare the efficiency of several derivative-free optimization methods in solving it, and show that our approach identifies the outliers correctly when ANNs are used for nonlinear regression.