Gaurav Bansal

An extended tuning method for cost-sensitive regression and forecasting

In many real-world regression and forecasting problems, over-prediction and under-prediction errors have different consequences and incur asymmetric costs. Such problems entail the use of cost-sensitive learning, which attempts to minimize the expected misprediction cost, rather than minimize a simple measure such as mean squared error. A method has been proposed recently for tuning a regular regression model post hoc so as to minimize the average misprediction cost under an asymmetric cost structure. In this paper, we build upon that method and propose an extended tuning method for cost-sensitive regression. The previous method becomes a special case of the method we propose. We apply the proposed method to loan charge-off forecasting, a cost-sensitive regression problem that has had a bearing on bank failures over the last few years. Empirical evaluation in the loan charge-off forecasting domain demonstrates that the method we have proposed can further lower the misprediction cost significantly.

Tuning Data Mining Methods for Cost-Sensitive Regression: A Study in Loan Charge-Off Forecasting

Real-world predictive data mining (classification or regression) problems are often cost sensitive, meaning that different types of prediction errors are not equally costly. While cost-sensitive learning methods for classification problems have been extensively studied recently, cost-sensitive regression has not been adequately addressed in the data mining literature yet. In this paper, we first advocate the use of average misprediction cost as a measure for assessing the performance of a cost-sensitive regression model. We then propose an efficient algorithm for tuning a regression model to further reduce its average misprediction cost. In contrast with previous statistical methods, which are tailored to particular cost functions, this algorithm can deal with any convex cost functions without modifying the underlying regression methods. We have evaluated the algorithm in bank loan charge-off forecasting, where underforecasting is considered much more costly than overforecasting. Our results show that the proposed algorithm significantly reduces the average misprediction costs of models learned with various base regression methods, such as linear regression, model tree, and neural network. The amount of cost reduction increases as the difference between the unit costs of the two types of errors (overprediction and underprediction) increases.

The role of privacy assurance mechanisms in building trust and the moderating role of privacy concern

Privacy policy statements and privacy assurance cues are among the most important website features that online providers use to increase individuals’ trust and willingness to disclose private information online. The focus of this study is a comprehensive examination of the process by which privacy assurance mechanisms influence trust and the moderating role of privacy concern in this process. We use the lens of the Elaboration Likelihood Model to investigate the way different individuals perceive and process privacy assurance mechanisms. We argue that the trust-enhancing role of these mechanisms depends on the individual’s privacy concern. The results of this study articulate the process by which various privacy assurance mechanisms operate in enhancing an individual’s trust, and show that there are distinct behavioral differences between individuals with high- vs low-privacy concern when forming their trust to disclose private information. The paper sheds new light on the role of elaboration in the trust building process, and shows why privacy assurance mechanisms have different impacts depending on individuals’ privacy concerns.

Emotional Intelligence (EI) and Nursing Leadership Styles Among Nurse Managers

Less than 12.5% of nurses aspire to leadership roles, noting lack of support and stress as major factors in their decision not to pursue this area of practice. Psychological resiliency, described as the ability to properly adapt to stress and adversity, is key to successful nurse managers. Emotional intelligence (EI) is a related concept to resiliency and is another noteworthy predictor of leadership and management success. This study was undertaken to determine the level of and relationship between EI and leadership style of nurse managers employed in Wisconsin and Illinois facilities. A descriptive, exploratory study design was utilized, with a convenience sample of nurse managers working in 6 large Midwestern health systems. Nurse managers were invited to participate in the study by their employer, completing the online consent form and the demographic, Multifactor Leadership Questionnaire (MLQ) Form 5X and the Emotional Quotient Inventory (EQ-i 2.0) surveys. Statistically significant positive relationships were noted between EI and transformational leadership and the outcomes of leadership (extra effort, effectiveness, and satisfaction). No statistically significant relationships were noted between EI and transactional or laissez-faire leadership styles.

E-Book Usage: Role of Environmental Consciousness, Personality and past Usage

E-books are a relatively new phenomenon and correspondingly, there is limited research on the factors associated with e-book usage especially in relation to printed books. In this research it is, therefore, of interest to examine antecedents of e-book usage intentions in one conceptual framework. The study was conducted using a course e-book. Findings reveal that environmental consciousness lowers the preference for printed books, and augments the design evaluation of the e-book. Personality factors impact the preference for printed books and perceived design evaluation of the e-book in different ways. A preference for printed books reduces, whereas perceived positive navigational design enhances, the attitude toward e-book usage. The findings have theoretical and practical implications. This study is one of the first to systematically study the e-book usage antecedents in a holistic manner, and also the first to examine the role of environmental consciousness in impacting e-book usage intentions in general.

Cultural Signifiers of Web Site Images

Web sites rely on pictures and animation to convey subtle messages that are more effectively communicated nonverbally. We argue that such messages could have strong cultural content, which should be understood in developing Web sites. Hence, this paper explores the cultural content of Web site images and develops a theory for Web-image signifiers. This is done in two phases. Phase I has an interpretive qualitative approach that uses Grounded Theory to identify signifiers and to develop the Web-image signifiers (WIS) theory. Phase II quantitatively tests the WIS theory. Together, these two phases identify and validate signifiers of cultural dimensions in Web site images. More interestingly, the results uncover that cultural dimensions are signified in five categories, of which two, humans and buildings categories, are the most prominent. The contribution of this paper is in developing a comprehensive theory for the cultural content of Web images, identifying 48 signifiers in Web images, discovering new categories of signifiers, and providing insights into the nature of cultural signification by testing the theory. Such knowledge could heighten our sensitivity and awareness of hidden cultural messages in Web site images. The WIS theory could provide a novel approach to the cultural studies of Web images and other artifacts with cultural content. The results of this work have immediate application in the design of Web sites for a multicultural audience.

Success Factors in Cooperative Online Marketplaces: Trust as the Social Capital and Value Generator in Vendors-Exchange Relationships

The business reliance on cooperative online exchanges for business-to-business transactions is on the rise. This paper examines the factors contributing to the success of vendor-exchange relationships in this type of marketplace. We use a Critical Realism approach to identify constructs salient to vendors-exchange relationships. A synthesis of value creation, social capital, and trust theories is used for conceptualizing the model. The model is tested using the data from a survey of vendors participating in a cooperative exchange. Results indicated that value creation was the main source of continuance in vendors-exchange relationship. However, the perception of value depended to a larger degree on the relationship quality and to a lesser degree on transactional exchanges (using the exchanges technology solutions as the sources of transaction), indicating the strength and extent of vendors-exchange relationship is an intangible asset for the exchange company. We also found that the exchange companys innovativeness was critical to the perceived quality of transactional exchanges, whereas perception of unfair treatment and communication quality influenced relationship quality. The results also indicated that trust was the main source for the social capital that contributed to perceived value of relationship. The ability of the exchange company to build a community was another factor contributing to the success of this relationship.

Distinguishing between Privacy and Security Concerns: An Empirical Examination and Scale Validation

ABSTRACT Even though privacy concerns (PC) and security concerns (SC) are acknowledged as separate constructs, many studies argue that they are related or, even worse, confused. Building upon the non-overlapping SC and PC dimensions, we propose an Internet Users Information Transmission Security Concern (IUITSC) scale. We argue that the scale when used in conjunction with the PC scale would cover the entire gamut of SCs and PCs (SC–PC). The construct is tested in a nomological network of personality and trust. The article first develops the theoretical foundation for multiple factor structures including various first-order factor models as well as the second-order factor model. The results, based on extensive statistical tests and nomological examination, suggest that the IUITSC may be represented as a second-order construct comprising of authentication, confidentiality, integrity, and non-repudiation. We argue that information assurance is the common theme that ties these constructs into a higher order IUITSC construct.

Data Breaches and Trust Rebuilding: Moderating Impact of Signaling of Corporate Social Responsibility

Data breach is a serious global security and trust concern. Data breaches have both direct and indirect as well as a short-term and long-term financial implications for the victim organization. One of the significant long-term financial costs is the loss of consumer trust in the organization. It seems that the sheer number and size of data breaches, insider data breaches, in particular, is not about to slow down anytime soon. There is a need to heighten information security to prevent such breaches, and there is a need to deploy trust-building strategies to minimize the trust fallout from such breaches. There is plenty of evidence that consumers use corporate social responsibility (CSR) as a means of differentiating one company from another. In this paper, we examine if CSR strategies such as embracing a social cause (LGBT and nature) could assist in repairing organizational trust in the wake of a data breach.