Solvency prediction for small and medium enterprises in banking

2017 | Decision Support Systems | Citations: 3

Authors: Figini, Silvia; Bonelli, Federico; Giovannini, Emanuele

Abstract: This paper describes novel approaches to predict default for SMEs. Multivariate ... Expand

Abstract: This paper describes novel approaches to predict default for SMEs. Multivariate outlier detection techniques based on Local Outlier Factor are proposed to improve the out of sample performance of parametric and non-parametric models for credit risk estimation. The models are tested on a real data set provided by UniCredit Bank. The results at hand confirm that our proposal improves the results in terms of predictive capability and support financial institutions to make decision. Single and ensemble models are compared and in particular, inside parametric models, the generalized extreme value regression model is proposed as a suitable competitor of the logistic regression. Collapse

Topics: logistics management missing data

Methods: logistic regression random forest classification machine learning ensemble learning regression analysis method

A Comparative Analysis of Univariate Time Series Methods for Estimating and Forecasting Daily Spam in United States

2016 | Americas Conference on Information Systems | Citations: 0

Authors: Zhang, Jie; Lee, Gene; Wang, Jingguo

Abstract: Spam has been one of the most difficult problems to be addressed since the inven ... Expand

Abstract: Spam has been one of the most difficult problems to be addressed since the invention of Internet. Outbound spam can reflect the information security level of an organization as most spam emails are generated by compromised computers. Understanding the trend of outbound spam can help organizations adopt proactive policies and measures toward a more informed decision on resource allocation in terms of security investment. This can also facilitate governments in designing incentive mechanisms such as cyber insurance. Using three years of daily spam volume from the United States, this research investigates the distribution of daily volume and compare univariate time series methods for forecasting. The methods explored include various exponential smoothing methods, auto-regressive integrated moving average (ARIMA), and neutral network auto-regressive (NNAR) modeling. Our evaluation of multiple univariate time series models suggests that Holt’s linear trend models based on exponential smoothing and neural network modeling can lead to more accurate spam prediction. The conclusion and implications are discussed. Collapse

Topics: email spam denial of service attack spamming computer crime electronic mail

Methods: time series analysis longitudinal research artificial neural network autoregressive integrated moving average descriptive statistic

An Empirical Analysis of Consumption Patterns for Mobile Apps and Web: A Multiple Discrete-Continuous Extreme Value Approach

2014 | International Conference on Information Systems | Citations: 0

Authors: Han, Sang Pil

Abstract: Using a unique panel data set detailing individual-level mobile app consumption, ... Expand

Abstract: Using a unique panel data set detailing individual-level mobile app consumption, this study develops a utility theory-based structural model for multiple discrete/continuous choices in app use. We identify the dynamics and interdependencies between mobile apps and jointly quantify consumers’ app choice and satiation simultaneously. The results suggest that mobile users’ baseline utility is the highest for communication apps, while the lowest for personal financing apps. In addition, users’ satiation level is the highest for the personal financing apps and the lowest for the game apps. However, a substantial heterogeneity in baseline utility and satiation is observed across diverse users. Furthermore, both positive and negative correlations exist in the baseline utility and satiation levels of mobile web and app categories. Consequently, the proposed frameworks could open new perspectives for handling large-scale, micro-level data, serving as important resources for big data analytics in general and mobile app analytics in particular. Collapse

Topics: mobile application big data social network mobile system Facebook

Methods: longitudinal research descriptive statistic design artifact econometric modeling extreme value analysis

Theories: utility theory

Auditing Journal Entries Using Extreme Value Theory

2013 | European Conference On Information Systems | Citations: 0

Authors: Argyrou, Argyris

Abstract: While a wealth of statutory and auditing pronouncements attest to the importance ... Expand

Abstract: While a wealth of statutory and auditing pronouncements attest to the importance of the auditing of journal entries for preventing and detecting material misstatements to financial statements, existing literature has so far paid inadequate attention to this line of research. To explore this line of research further, this paper proposes a bipartite model that is based on extreme value theory and Bayesian analysis of Poisson distributions. The paper assesses the veracity of the model via a series of experiments on a dataset that contains the journal entries of an international shipping company for fiscal years 2006 and 2007. Empirical results suggest the model can detect journal entries that have a low probability of occurring and a monetary amount large enough to cause financial statements to be materially misstated. Further investigations reveal that the model can assist auditors to form expectations about the journal entries thus detected as well as update their expectations based on new data. The findings indicate that the model can be applied for the auditing of journal entries, and thus supplement existing procedures. Collapse

Topics: accounting database system

Methods: extreme value analysis anomaly detection Bayesian analysis reference modelling experiment

Developing a decision support system for business model design

2013 | Electronic Markets | Citations: 17

Authors: Daas, Dave; Hurkmans, Toine; Overbeek, Sietse; Bouwman, Harry

Abstract: Software vendors increasingly offer web-based services (SaaS) at a subscription ... Expand

Abstract: Software vendors increasingly offer web-based services (SaaS) at a subscription fee. Because they need a network of service partners to generate customer value, SaaS providers have to focus on business model (BM) design. In this study, a decision support system (DSS) is developed to help SaaS in this process, based on a design approach consisting of a design process that is guided by various design methods. The outcome of the design process is a computer-based tool that takes important BM evaluation criteria, such as critical design issues and success factors, into account, by integrating market analysis based on conjoint analysis, providing alternative designs for business models involving companies working together in a value network offering e-services, and including a multi-criteria decision analysis, based on AHP, that can be used to select the business model of the core services and identify the implications for the business model of all the actors involved in the network. The DSS is discussed in an illustrative case. The execution of the design process provides insight into how BM design processes can be improved and formalized. The BM decision-making tool illustrates the importance of modularity, i.e. the BM-DSS can be divided into several independent sub-components, while at the same time maintaining its usefulness within an organizational network context. Collapse

Topics: decision support system software as a service business model price management willingness to pay

Methods: case study design framework qualitative interview design process design methodology

A Value-at-Risk Approach to Information Security Investment

2008 | Information Systems Research | Citations: 0

Authors: Wang, Jingguo; Chaudhury, Aby; Rao, H. Raghav

Abstract: Information security investment has been getting increasing attention in recent ... Expand

Abstract: Information security investment has been getting increasing attention in recent years. Various methods have been proposed to determine the effective level of security investment. However, traditional expected value methods (such as annual loss expectancy) cannot fully characterize the information security risk confronted by organizations, considering some extremal yet perhaps relatively rare cases in which a security failure may be critical and cause high losses. In this research note we introduce the concept of value-at-risk to measure the risk of daily losses an organization faces due to security exploits and use extreme value analysis to quantitatively estimate the value at risk. We collect a set of internal daily activity data from a large financial institution in the northeast United States and then simulate its daily losses with information based on data snapshots and interviews with security managers at the institution. We illustrate our methods using these simulated daily losses. With this approach, decision makers can make a proper investment choice based on their own risk preference instead of pursuing a solution that minimizes only the expected cost. Collapse

Topics: data security website security policy IT security enterprise information system

Methods: extreme value analysis augmented Dickey–Fuller test qualitative interview sensitivity analysis descriptive statistic

Theories: decision theory

An Extreme Value Approach to Information Technology Security Investment

2005 | International Conference on Information Systems | Citations: 0

Authors: Wang, Jingguo; Chaudhury, Abhijit; Rao, Raghav

Abstract: Information technology security investment is receiving increasing attention in ... Expand

Abstract: Information technology security investment is receiving increasing attention in recent years. Various methods have been proposed to determine the effective level of security investment. In this paper, we introduce an extreme value approach to address the issues of effective budgeting and investing in IT security. In our model, the security status of a system depends on two factors: system security level, which is measured by the level of security investment, and system attack level, which reflects the security risk with which the system is confronted. Security investment level is endogenous to the system, while attack level is exogenous. Extreme value analysis is used to characterize the stochastic behavior of high-level attacks based on the historical data and to make inferences on future attacks. Based on these inferences, we determine the effective security solutions and the level of security investment to modulate the likelihood of system failure. For illustration purposes, we use an extreme value approach to analyze a set of traffic data collected from a regional bank. Collapse

Topics: IT security IS failure email spam internet technology electronic business

Methods: extreme value analysis regression analysis method sensitivity analysis survival function literature study