Abstract

We investigate an unexplored channel—loss given default (LGD)—through which accounting information can shape the design of debt contracts. Using a sample of defaulted bonds, we find that borrower accounting information available at contract initiation possesses significant power for predicting realized LGD at the subsequent default date. We then use this model to construct an accounting-based measure of expected LGD at the contracting date for a large sample of bond issuances. We find that this measure is positively associated with issuance date interest spread and covenant use, and document that these relations are not artifacts of an association between LGD and probability of default. We then show that accounting-based expected LGD has a stronger association with issuance date spread when the borrower’s underlying accounting is more conservative and when the accounting-based LGD predictors are more persistent. Our results increase our understanding of both the informational role and contracting role of accounting information.

Introduction

Financial reports are an important source of frm-specifc information available to lenders at the contracting date and thus may afect lenders’ behavior in the design of debt contracts. Whereas prior literature focuses on the usefulness of accounting information for probability of default assessment (e.g., Beaver 1966; Altman 1968; Shumway 2001), this study suggests that accounting information may also shape contracts by facilitating lenders’ assessment of loss given default. Loss given default, defned as the percentage loss that lenders experience from $1 of outstanding principal in a case of default, is a critical component of credit risk and debt contracting theories.1 Despite the theoretical importance of loss given default, there is little empirical evidence regarding how accounting information available to lenders at the contracting date is associated with their expectations about loss given default. This study is the frst to provide evidence that accounting information at the contracting date is a useful predictor of realized loss given default and that lenders behave as if they use accounting information about loss given default to design debt contracts.

Conclusion

A substantial body of research focuses on the informativeness of accounting information with respect to probability of default. However, there is very little research concerning the informativeness of accounting information with respect to loss given default, particularly at the contracting date. There exists some ancillary evidence in extant literature regarding the informativeness of accounting data about loss given default at the time of default. However, given the relatively long time between debt issuance and eventual default, this evidence says little about whether accounting information is useful to lenders in assessing expected loss given default at the contracting date—when lenders need information the most.

This study contributes to the literature along several dimensions. First, using a sample of defaulted bonds, we show that accounting measures available to lenders at the contracting date are informative about future loss given default. To the best of our knowledge, this study is the frst to do so. This fnding complements the literature which shows that accounting measures are informative about probability of default and therefore enhances our understanding of the informational role of accounting. Second, we construct an intuitive measure of accounting-based expected loss given default at the time of bond contract initiation, which could be of use in future research. Third, we show that lenders behave as if their accounting-based expectations about loss given default signifcantly afect price and non-price terms of bond contracts. We provide evidence that the association is stronger when the accounting-based predictors are more persistent, consistent with the intuition that accounting information with higher persistence is more reliable for predicting future outcomes. This fnding contributes to the literature on the role of accounting information in debt contracting by showing a specifc channel through which accounting information may be useful in lending decisions. Finally, our results provide important evidence that establishes a direct link between accounting-based LGD estimation and lenders’ demand for fnancial reporting conservatism.

Abstract

     Depression is a serious mental health condition that may lead to poor mental and emotional functioning at work, at school and in the family causing the mental imbalance. In worst scenarios, depression may lead to severe anxiety or suicide. Hence, it is necessary to diagnose depression at early stages. This paper elaborates the development of a novel approach for a convolutional neural network model that can examine facial images from the recorded interview sessions to discover facial patterns that could indicate depression level. The user-generated data helps to distinguish between different depressive groups with depression symptoms that can manifest people with various mental illnesses in different ways. In particular, we want to automatically predict the depression scale and differentiate depression from other mental disorders using the patient’s psychiatric illness history and dynamic textual descriptions extracted from the user inputs. We apply the k-nearest neighbour algorithm on the dynamic textual descriptors to make a linguistic analysis for classifying mental illness into different classes. We apply dimensionality reduction and regression using the Random Forest algorithm to predict the depression scale. The proposed framework is an extension to pre-existing frameworks, replacing the handcrafted feature extraction technique with the deep feature extraction. The model performs 2.7% better than existing frameworks in facial detection and feature extraction.

Introduction

     In India, depression is common in adults between the ages of 16–25, considered to be a leading cause of disability. For years, researchers have to identify and map the relationship between brain function and structure using neuroimaging data. The researchers at the University of Texas have identified a unique technique to categorize people susceptible to developing depression and anxiety using deep learning with a supercomputer. Now they are using the Stampede supercomputer at the Texas advanced computing centre to train deep learning algorithms that can identify similarities among hundreds of patients using magnetic resonance imaging genomics data and other factors to predict patients at risk of depression and psychological disorders (Valstar et al., 2013). In the past, the researchers have worked on the development of a model that takes raw text and audio segments as input and analyzes the wave-forms that predict depression. The work provides a method for deep learning-based segmentation to detect depression, as well as irregular segmentation and masks used for Gabor wavelength detection (Long et al, 2014). A Gabor filter, named after Dennis Gabor, is a linear local texture filter in image processing. It analyses if the model includes any specific frequency content in specific directions in a localized region around the point or region of analysis. The researchers trained the deep learning algorithm using extracted audio and textual features from clinical patients with suicide attempts and higher rate of mood swings (Ma et al, 2016). Extreme Gradient boosting technique is used for identifying and categorizing the important parameters of depression and predicting depression cases by re-sampling methods using different balanced samples but the researchers were not able to precisely predict the depression scale. Also differentiating the mental disorders is a tedious task due to similarities in symptoms (Marcus et al., 2012).

Conclusion and future work

     Here, we have developed an AI-based automated system for predicting the depression scale. The model is purely based on recognizing facial expressions from recorded video sessions of depressed patients. We have used deep learning algorithms for extracting dynamic facial features and recognizing expressions producing the feature dynamic history histogram applied to feature dynamic vector sequences applied to the video sample. After calculating the feature dynamic history histogram, we apply regression techniques to correlate between feature and depression scales. The experimental results showed that the model had performed well on the test video dataset. The deep feature extraction technique significantly performs much better than the handcrafted feature extraction technique. The feature attributes are directly extracted from the responses provided by the convolutional layers perform best rather the performance of the neural networks are closely connected giving benchmarking results.

The malware detection methods are classified into two categories, namely, dynamic analysis (active analysis) and static analysis (passive analysis). These methods undergo unusual obstruction, and challenges that are process complexity, limitation over detection accuracy. The static method serves to discover malicious applications using various parameters like permission analysis, signature verification. It can be regularly obfuscated. Dynamic techniques entail investigating the performance of an application by administering it in a restricted environment. The complex version of a portable executable often emerges with an intervention by hardening the dynamic analysis centric malware detection methods. The various constraints of these dynamic and static models contribute to this manuscript represents a Multi-Level Malware detection using Triad Scale (MLMTS) built on regression coefficients. The proposed method MLMTS spans into three levels, such that the first and second level performs static analysis, and the third level performs the dynamic analysis. The second and third levels of the hierarchy invoke upon the ambiguous decision of their respective predecessor level. The proposed work is based on the Machine Learning (ML) model that determines the triad scale by applying linear regression for each level of malware detection. The call sequences of the portable executable, arguments passed to these call sequences and their fallouts (resultant values) in respective order of three levels of the MLMTS method. The experimental study manifests the significance of the proposal compared to the other recent malware detection methods.

1. Introduction

The malicious software that is often termed as malware intends to infiltrate, infect, or intrude the cryptographic verification of the owner in the computer system. According to contemporary statistics [1], an average of 400 million malware models is recognized per annum. Currently, the malware family has boosted through the software modules engineered by incredible software skills [2]. The attacks can anonymize the source of the attack [2] and considerably succeeds in hacking the potential industrial structures known as the Stuxnet [3]. Anonymized sources are significant challenges to contemporary malware detection strategies. Extensive utilization of computer-networks is vulnerable to potential malware attacks. The dynamic network connectivity exposes the vulnerabilities of the corresponding network, which entertains the attackers to exploit these vulnerabilities to inject the malware into the respective system. The Intel organization has estimated the impact of malware in terms of loss of revenue as above 400 billion (USD 400 * 109 ) dollars worldwide per annum [1]. These statistics concreting the need for potential malware detect and defense mechanisms.

1- Introduction

2- Related study

3- Methodology

4- Experimental result

5- Discussion

6- Conclusion and future work

References

Abstract

Diabetes causes a large number of deaths each year and a large number of people living with the disease do not realize their health condition early enough. In this study, we propose a data mining based model for early diagnosis and prediction of diabetes using the Pima Indians Diabetes dataset. Although K-means is simple and can be used for a wide variety of data types, it is quite sensitive to initial positions of cluster centers which determine the final cluster result, which either provides a sufficient and efficiently clustered dataset for the logistic regression model, or gives a lesser amount of data as a result of incorrect clustering of the original dataset, thereby limiting the performance of the logistic regression model. Our main goal was to determine ways of improving the k-means clustering and logistic regression accuracy result. Our model comprises of PCA (principal component analysis), k-means and logistic regression algorithm. Experimental results show that PCA enhanced the k-means clustering algorithm and logistic regression classifier accuracy versus the result of other published studies, with a k-means output of 25 more correctly classified data, and a logistic regression accuracy of 1.98% higher. As such, the model is shown to be useful for automatically predicting diabetes using patient electronic health records data. A further experiment with a new dataset showed the applicability of our model for the predication of diabetes.

Introduction

Diabetes stands among the top 10 causes of death for 2016. Diabetes killed 1.6 million people in 2016, up from less than 1 million in 2000. With this figure diabetes replaced HIV/AIDS as the seventh top cause of death [1]. The number of people with diabetes has risen from 108 million in 1980 to 422 million in 2014, with the global prevalence of diabetes among adults over 18 years of age rising from 4.7% in 1980 to 8.5% in 2014 [2]. By 2040, 642 million adults (1 in 10 adults) are expected to have diabetes. Also, 46.5% of those with diabetes have not been diagnosed [3]. In order to reduce the number of deaths attributable to diabetes, it is essential that methods and techniques that will aid in early diagnosis of diabetes be devised, because a large number of deaths in diabetic patients are due to late diagnosis. In order to achieve cutting-edge techniques for the early diagnosis of diabetes, we need to utilize advanced information technology, and data mining is a suitable field for this. Data mining offers the ability to extract and discover previously unknown, hidden, but interesting patterns from a large database repository. These patterns can aid medical diagnosis and decision-making. Various techniques and algorithms have been designed for application in extracting knowledge and information in the diagnosis and treatment of disease from medical databases. PCA is a simple, nonparametric method for extracting relevant information from confusing data sets [4]. When a large dataset is to be clustered into a user specified number of clusters (k), which are represented by their centroids, k-means will cluster the data by minimizing the squared error function [5], and often misclassifies some data due to outliers; also the time complexity will be greater. To overcome these problems, principal components analysis (PCA) can be used to reduce the dataset to a lower dimension, while ensuring that the least information is lost, and providing a better centroid point for clustering. K-means clustering partitions a dataset into different groups of similar objects. Clusters that are highly dissimilar from the others are regarded as outliers and discarded. Logistic regression is an efficient regression predictive analysis algorithm. Its application is efficient when the dependent variable of a dataset is dichotomous (binary).

1- Introduction

2- Finite element modeling and artificial data generating

3- Variable selection and parametric study

4- Determine B/D ratio for overhang construction

5- Validation

6- Conclusion

References

Abstract

Wide flange beams are widely used in the United States for bridge design and construction. During the overhang construction of the bridge, torsional loads are often induced due to the fresh concrete load and the use of a deck finishing machine located on the overhang formwork. These torsional moments sometimes cause excessive exterior girder rotation, resulting in many safety and maintenance issues during the construction and service stages. To prevent these issues, most states have specifications for limiting the rotation. Finite element analysis using shell or solid elements is usually recommended for analyzing bridge girders in overhang construction, which can be tedious and difficult in some cases. This study focused on developing a simple method with minimal calculation to evaluate the ratio of unbraced length to girder depth (B/D ratio). The stepwise variable selection method and a regression analysis were conducted to find the relationship between the exterior girder rotation and bridge geometries. A computer program for automatic finite element modeling in SAP2000 was developed using MATLAB, resulting in 4285 finite element models with different bridge geometries being developed to generate artificial data. By conducting a study of variable selection, three parameters were selected based on level of significance. After conducting a regression analysis based on the selected parameters, a method using the normal weight of girder, overhang width, and rotation limit to determine B/D ratio was developed.

Introduction

In the U.S., wide flange beams are commonly used as supports for a bridge deck slab and traffic loads. Typically, these wide flange beams are uniformly spaced transversely, with the bridge deck overhang cantilevering past the exterior girder. The construction of the deck overhang often requires utilizing overhang brackets to support the fresh concrete, deck finishing machine, and other construction loads. As shown in Fig. 1, deck overhang brackets are generally connected to the top flange and react against the bottom of the exterior girders’ web, with spacings between 0.9 m (3ft) to 1.2 m (4ft) along the exterior girders [1,2]. The deck finishing machine, generally located on the edge of the overhang, creates significant loads on the bridge’s exterior girders during deck construction (Fig. 2). One of the major issues in deck construction is unbalanced loads applied along the overhang portion of the deck slab, mostly due to the weight of the deck finishing machine and fresh concrete [3–5]. During deck construction, the bridge shows less transverse stiffness since the concrete deck slab has yet to provide stiffness to the structure. Also, the construction loads are transferred through the overhang bracket formwork system to the exterior girder, causing significant torsional moments on the exterior girder. These torsional moments can lead to excessive transverse rotation of the exterior girder, and instigate many safety and maintenance issues during both the construction and service stage of the bridge, such as changes in deck thickness [3,6], local and global instabilities [7–10], and potential bridge failure [11–13]. Therefore, AASHTO LRFD Bridge Design Specification (2012) [14] requires that the effect of reactions from the overhang brackets be considered during bridge design, and recommends using a three-dimensional (3-D) finite element analysis using shell or solid elements for the torsional analysis in order to recognize the warping effects. The Guidelines for Steel Girder Bridge Analysis (2011) [15] introduces a method using the equivalent torsional constant, which includes the St. Venant torsional stiffness and warping fixity at each end of a given unbraced length, to determine the rotation of the girders.

Majority of conventional prediction methods are based on the assumption of linear dependence of the predicted (dependent) variable and the set of independent descriptors. The multiple linear regression models are built using the correlation and regression analysis methods. Nonlinear relationships between the variables are conventionally modeled via fi tting of nonlinear curves (quadratic, cubic, power, exponential, logarithmic, hyperbolic, or logistic ones) or their linearization. However, the artifi cial neural networks approach has been recognized as promising method to model nonlinear dependences in the prediction tasks over recent decades [1]. Major advantages of the artifi cial neural networks algorithms are their capability to learning, generalization, and prediction of the data, fault tolerance, parallel data processing, and fast computation procedures. This has been supported by the neural networks application in theoretical and computational chemistry, analytical chemistry, biochemistry, medicine, drugs chemistry, pharmaceutics, and food products studies. It should be noted that artifi cial neural networks have been applied to the chemometrics tasks since early 1990ies. Four applications of artifi cial neural networks in chemical engineering have been comprehensively described [2]: fault detection, quality prediction, signal processing, and modeling and control of the processes. Various architectures of artifi cial neural networks and their applications in chemistry have been demonstrated [3], outlining the advantages and disadvantages in comparison with conventional chemometrics methods. A novel approach to prediction of biological activity of peptides and proteins, physics and chemistry-driven artifi cial neural network (Phys-Chem ANN), has been proposed [4]. The Phys-Chem ANN has been based on physical and chemical properties as well as structural features of proteins. The task on classifi cation and prediction of the strength of weak organic acids in aqueous-organic solvents has been solved [5].

1- Introduction

2- Bayesian nonparametric regression through Gaussian process prior

3- Application

4- Results

5- Concluding remarks

References

Abstract

The goal of this paper is to promote the use of Non-Parametric Regression (NPR) for hypothesis testing in hospitality and tourism research. In contrast to linear regression models, NPR frees researchers from the need to impose a priori specification on functional forms, thus allowing more flexibility and less vulnerability to misspecification problems. Importantly, we discuss in this paper a Bayesian approach to NPR using a Gaussian Process Prior (GPP). We illustrate the advantages of this method using an interesting application on internationalization and hotel performance. Specifically, we show how in contrast to linear regression, NPR decreases the risk of making incorrect hypothesis statements by revealing the true and full relationship between the variables of interest.

Introduction

Despite the increased popularity of non-parametric regression (NPR), its use in the tourism and hospitality literature remains very limited. We aim in this note to highlight the advantages of NPR, and illustrate how it can be used to provide a more accurate reflection on the true relationship between a set of variables. We show through an example that hospitality researchers might be missing some important input for hypothesis testing when estimating the traditional linear regression model. NPR, like linear regression, estimates mean outcomes for a given set of covariates. However, unlike linear regression, NPR is not subject to misspecification error arising from potentially wrong functional forms as it does not impose a priori a functional form on the regression model (Müller, 2012; Mammen et al., 2012). The linear model (y = β0 + βx + u) is generally assumed for convenience, and not because we truly believe that the model is linear in reality. Researchers in the field often model nonlinearities using extensions of the linear model, for example, y = β0 + β1x + β2×2 + u. It is clear, however, that this model accounts only for limited types of nonlinearity of U or inverted U shape, and cannot capture more complicated patterns in the data. When more than one regressor is available, nonlinearities are often modeled using interactions: y = β0 + β1x + β2z + β3xz + u. The interpretation is that the effect of x on y depends on z: = + ∂ ∂ β β z E y x ( ) 1 3 . This is, of course, a deviation from the simple linear model where the main assumption is that the effect of x on y is constant across all values of x or other explanatory variables. However, the effect of x on y depends on z in a linear way, an assumption that may or may not hold in practice.

This research introduces the Recursive General Regression Neural Network Oracle (R-GRNN Oracle) and is demonstrated on several binary classification datasets. The traditional GRNN Oracle classifier (Masters et al., 1998) combines the predictive powers of several machine learning classifiers by weighing the amount of error each classifier has on the final predictions. Each classifier is assigned a weight based on the percentage of errors it contributes to the final predictions as the classifiers evaluate the dataset. The proposed R-GRNN Oracle is an enhancement to the GRNN Oracle in which the proposed algorithm consists of an oracle within an oracle – where the inner oracle acts as a classifier with its own predictions and error contribution. By combining the inner oracle with other classifiers, the R-GRNN Oracle produces superior results. The classifiers considered in this study are: Support Vector Machine (SVM), Multilayer Perceptron (MLP), Probabilistic Neural Network (PNN), Gaussian Naïve Bayes (GNB), K-Nearest Neighbor (KNN), and Random Forest (RF). To demonstrate the effectiveness of the proposed approach, several datasets were used, with the primary one being the publicly available Spambase dataset. The predictions of SVM, MLP, KNN, and RF were used to create the first GRNN Oracle, which was then enhanced with the high performances of SVM and RF to create the second oracle, the R-GRNN Oracle. The combined recursive model was 93.24% accurate using 10-fold cross validation, higher than the 91.94% of the inner GRNN Oracle and the 91.29% achieved by RF, the highest performance by a stand-alone classifier. The R-GRNN Oracle was not only the most accurate, but it also had the highest AUC, sensitivity, specificity, precision, and F1-score (97.99%, 91.86%, 94.40%, 93.28%, and 92.57%, respectively). The research contribution of this paper is introducing the concept of recursion (a concept not fully explored in machine learning models and applications) and testing this structure’s ability on further enhancing the performance of the traditional oracle. The recursive model has also been applied to several other datasets: The Human Resources, Bank Marketing, and Monoclonal Gammopathy of Undetermined Significance (MGUS) datasets. The results of these implementations are summarized in this paper.

Introduction

The world today produces huge and complex amounts of data every second. In this new era of big data, advanced analytic methods can extract valuable information, patterns, trends, and associations to provide meaningful insights. Processing such data manually would be impractical if not impossible, therefore, the need to automate such processes is needed. Tasks too complex for humans to code and process directly require machine learning. Machine learning helps analyze big data by focusing on designing algorithms that can learn patterns in the data to make predictions. It is a branch of artificial intelligence that teaches machines how to learn from experiences and adapt. Successful data mining requires effective machine learning techniques. Data mining is defined as the process of discovering properties and extracting valuable information from large, incomplete, and noisy raw data that is stored in databases, data warehouses, or other information repositories. In data mining, the data is stored electronically and is processed through computers (Witten, Frank, Hall, & Pal, 2016). It is about solving problems by analyzing data already present in databases, where some of its tasks include association rule learning, clustering, classification, and regression (Esfandiari, Babavalian, Moghadam, & Tabar, 2014). It is applied to various disciplines and industries such as manufacturing, customer relationship management, fraud detection, banking, marketing, and healthcare. The General Regression Neural Network Oracle (GRNN Oracle), developed by Masters et al. in 1998, combines the predictions of individually trained classifiers and outputs one superior prediction.

Introduction

As knowledge-based economies highlight the importance of innovation and advanced knowledge and skills, many firms have been engaging in innovation activities to achieve high performance (Audretsch et al. 2014; Coad and Rao 2008; Coad et al. 2016; Colombelli et al. 2013; Del Monte and Papagni 2003; Garcı´a-Manjo´n and Romero-Merino 2012). Innovation activities are the primary driving forces for sustainable growth because innovation helps firms achieve superior performance by developing new products or processes (Roper et al. 2008; Teece 1986; Wang and Thornhill 2010), and absorptive capacity, a byproduct of innovation activity, helps firms commercialize new information and knowledge (Bellucci and Pennacchio 2015; Cohen and Levinthal 1990; Wang and Thornhill 2010). One important effect of innovation is to deter the entry of new firms by creating entry barriers (Lukach et al. 2007; Phillips 1971), and this clearly reduces market competition (Del Monte and Papagni 2003; Phillips 1971). Therefore, firms aim to achieve a competitive advantage through innovation activities. However, converting innovative activities into firm performance takes considerable time and resources, and many difficulties arise in overcoming uncertainty at every stage of the innovation process (Coad and Rao 2008). The effect of innovation activities on firm growth often depends on specific unobservable factors (Audretsch et al. 2014; Garcı´aManjo´n and Romero-Merino 2012), and there is a lack of understanding of the relationship between innovation activities and the expected future performance of a firm. Specifically, scholars have yet to examine the underlying mechanism linking firms’ innovation activities, firm growth, and research and development (R&D) investment growth based on a comprehensive perspective. Although previous literature has revealed the importance of innovation for firm growth (Coad and Rao 2008; Colombelli et al. 2013), it has not explored the mechanism connecting innovation activities with R&D investment through firm growth. It is unclear where firms should invest for their growth, although one potentially important channel for investment is R&D, a critical input to innovation and growth in knowledge-based economies (Brown et al. 2009). Moreover, although R&D investment is important for sustainable growth, it appears to be a double-edged sword (Coad et al. 2016): the uncertainty and long payback period of innovation activities might reduce firm growth, and this concern can eventually lead firms to discourage investment in R&D. While numerous government-level studies have been conducted on topics such as R&D subsidies and fiscal treatments, firm-level studies on investment in R&D are still lacking (Coad and Rao 2010). In addition, given that small and medium-sized enterprises (SMEs) have been gradually expanding their role in the national economy, the differences between SMEs and large companies in terms of innovation activities and growth outcomes have not received enough academic attention. Examining the complicated relationship between innovation activities and firm growth is important not only to scholars but also to managers and investors who have to allocate their limited resources to firms’ R&D (Oriani and Sobrero 2008).

In this paper, we study and analyze the regularized least squares for functional linear regression model. The approach is to use the reproducing kernel Hilbert space framework and the integral operators. We show with a more general and realistic assumption on the reproducing kernel and input data statistics that the rate of excess prediction risk by the regularized least squares is minimax optimal.

Introduction

There are increasing cases in practice where the data are collected in the form of random functions or curves. This type of data is becoming more prevalent throughout science, engineering and financial market, as automated on-line data collection facilities are becoming more ubiquitous. Many classical statistical tools and models for multivariate analysis, such as principal components analysis, canonical correlation analysis and linear model are then extended to the infinite-dimensional functional domain. In this paper, we consider the functional linear model where Y is a scalar response, X : I → R is a square integrable functional predictor defined over compact domain I ⊂ R, α0 is the intercept, β0 : I → R is the slope function. and  is the random noise with mean 0 and finite variance σ 2 . Functional linear model was introduced by J.O. Ramsay and C.J. Dalzell [11] and first written in its commonly encountered form (1) by T. Hastie and C. Mallows [7]. Some recent research on the statistical analysis of (1) includes [2, 3, 6, 15, 16]. In this paper, we focus on the random design where X is a path of a square integrable stochastic process defined over I and is independent of . Without loss of much generality, throughout the paper we assume E(X) = 0 and the intercept α0 = 0, since the intercept can be easily estimated. Let L 2 be the Hilbert space of square integrable functions on I (with respect to Lebesgue measure) with standard inner production < u, v >= R I u(s)v(s)ds and norm kuk = R I u 2 (s)ds
1/.

Introduction

Theoretical review

Referent group factor

Intrinsic interest factor

Accountants’ work

Professional prestige factor

Methodology

Results

Conclusions

References

Abstract

Purpose: The purpose of this paper is to examine the attitude of business students towards the accounting profession and investigate the relationship between students’ attitude and their intention to pursue a degree in accounting.

Design/methodology/approach: A self-administered survey was used to collect data from students from the University of Ghana Business School (UGBS). A total of 457 questionnaires were used in the empirical analysis. A binary logistic regression analysis technique was employed to analyse the data.

Findings: The logistic regression analysis demonstrates that intrinsic interest in the accounting discipline, prior exposure to accounting at the senior high level and the desire to pursue professional accounting qualification in future are good predictors of students’ intention to major in accounting. The results also indicate family members, course instructors and other referent group play a crucial role in influencing students’ intention to pursue a career in accounting.

Introduction

One topical issue that has received considerable interest in accounting research, particularly in the field of accounting education is the shortages in the supply of professional accountants to meet the ever-growing demand. Prior studies have shown that the demand for professional accountants in many parts of the world, have over the years, exceeded the supply (Chen, 2012; Nor et al., 2013; Van Zyl and de Villiers, 2011). Interestingly, while countries are still struggling to deal with the shortages in the supply of professional accountants, it is expected that the demand for accountants globally will increase in the future due to the rapid growth in business activities (Coe, 2016). By implication, the challenge of meeting the demand of accountants globally could worsen if urgent steps are not taken to address the deficit in the supply of accountants. While the reasons for the global shortage in the supply of accountants remain an open question, a common theme emerging from prior empirical works attribute the unfortunate trend to the continuous decline in the number of university students who pursue accounting as a course major and the unwillingness of some accounting graduates to pursue accounting as a career ( Jackling and Calero, 2006; Cappelletto, 2010). In part, the negative perceptions students have about the accounting profession have been acknowledged to be an important source of this problem (Wells and Fieger, 2006). As pointed out in some studies (Baxter and Kavanagh, 2013; Hiltebeitel et al., 2000), young people usually develop their career aspirations based on preconceived ideas, inaccurate perceptions about professions and their work environment. In the field of accounting, in particular, a number of studies (Baxter and Kavanagh, 2013; Byrne and Willis, 2003; Jackling, 2001) document that students who are usually not attracted to the accounting field perceive it to be uninteresting, a discipline that requires good numeracy skills, full of rule memorising and a profession of boring people who rarely work among others. Thus, stereotypical image of accounting has been recognised to be a key contributing factor to the decline in the number of students pursuing accounting as a course major in schools. Accordingly, a number of studies seeking to examine how students perceive the accounting profession and the effect of such perceptions on their choice of academic major and career decisions have emerged (Albrecht and Sack, 2000; Baxter and Kavanagh, 2013; Byrne and Willis, 2003; Jackling and Calero, 2006; Tang and Seng, 2016). In particular most of these studies have explored the dominant factors that influence students’ decision to pursue a career in accounting.

In economics teaching, traditional lecturing still takes up the largest share of class time, an estimated 60% (Goffe and Kauper, 2014) to 83% (Watts and Schaur, 2011). This central role of lecturing is under increasing scrutiny as empirical evidence suggests that active learning techniques are more effective than lecturing in promoting learning (see e.g. Freeman et al., 2014). A pedagogical approach that decreases lecturing, thus freeing class time for active learning, is the inverted classroom (Lage et al., 2000) or the classroom flip (Baker, 2000), whereby first exposure to the material is moved outside the classroom usually in the form of lecture videos (Abeysekera and Dawson, 2015). The classroom flip appears to improve learning outcomes as reported in three major reviews of the existing literature (Bishop and Verleger, 2013, Giannakos et al., 2014, O’Flaherty and Phillips, 2015, O’Flaherty et al., 2015) even though reviewers express some concern for the lack of a “robust scientific approach” in evaluating these learning outcomes (O’Flaherty and Phillips, 2015, 89). Recent studies using more robust methods, seem to confirm the existence of improvement in learning outcomes from flipping the classroom, albeit moderate (Anderson and Brennan, 2015, Calimeris and Sauer, 2015). Moving beyond the analysis of how flipping the classroom affects learning outcomes overall, some researchers have begun to investigate more fine-grained questions. Ryan and Reid (2016) asked how flipping the classroom affects the outcomes of weaker students. They found a 56% reduction in Ds and Fs grades and in the withdrawal rate when flipping the classroom even though no improvements in learning emerged at the aggregate level. Touchton (2015) focused on what type of learning flipping affects the most. He found a larger improvement in learning outcomes in the sections which students generally find most challenging even though at the aggregate level the improvements were very small. Olitsky and Cosgrove (2016) examined whether gains in learning outcomes become larger as students adapted to the flip and showed that this was the case: the gains in learning increased as the flipped course progressed and students became better acquainted with the approach. Jensen et al. (2015) raised the question of how much the impact of the flipped classroom on learning outcomes depends on the choice of the control against which the learning outcomes of the flip are evaluated. They found insignificant learning benefits of the flipped classroom compared with a non-flipped, active learning course

Geotechnical engineers face a number of uncertainties [1, 2]. Soil materials formed from geological weathering processes, and by physical means to deliver the soil to the current position [3]. In the forming process, the soil is influenced by various stress, pore fluid, and physical and chemical changes. Therefore, it is not surprising that there are always some outliers in geotechnical data. When dealing with geotechnical problems, empirical correlations between in situ or laboratory test results and geotechnical parameters are often used in geotechnical design. When establishing such empirical correlations, mostly adopted method is regression analysis, including linear or nonlinear regression [4]. Linear least squares regression (LLR) is a modeling approach by far the most widely used. When people say they use “regression”, “linear regression” or “least squares” to adapt their data, they usually mean doing LLR. LLR is not only the most widely used method of modeling, but also have adapted to a variety of circumstances, beyond its immediate scope [5]. A mathematical method that finds the best-fit curve for a given set of points is to minimizing the sum of the squares of the distances of regression data deviating from the curve. The sum of squares of the offset distances is used instead of the absolute values of the offset distances because this allows the residuals to be treated as a continuously differentiable quantity. Whereas, because of the use of the square of the offset, peripheral points may have a disproportionate effect on fit. Whether the results are desirable or not, it depends on the issue of question [6]. The statistical observations of outliers are significantly different from the other sample values. Least-squares regression is obviously the best option if errors are normally distributed. Then, other means is eagerly required if these errors are not normally distributed. One particular distribution is the long tail error distribution of great concern. One solution is still to use the LLR method after removing the largest remaining value as outliers. However, this solution may be infeasible if several larger residual values exist by reason that the poor nature of the outlier tests. In addition, outlier testing is an acceptance/rejection process. The testing processes are neither smooth or statistically efficient. Robust regression (RR) is another option for least-squares regression in the case of the data contaminated with outliers. It can also be used to detect influential observations when the data is exposed to outliers [7]. It is difficult to define absolute outliers from geotechnical testing data, but it is possible to indicate the least predictable or relatively outlying data points using statistical tools. The objective of this paper is to demonstrate the advantages of RR analysis used in geotechnical data comparing with least square regression analysis. The procedure of RR is discussed shortly, based on LLR method. And then, Regression analysis is operated on undrained shear strength (su) data derived from CPTU test with both RR and LLR. Comparing regression result, the outlier of su data can be detected based on RR method.