Beyond Binary Boundaries – Exploring the Potential of Categorical Variables in Counterfactual Explanations

2023 | International Conference on Business Informatics | Citations: 0

Authors: Hühn, Philipp; Hofer, Simon

Abstract: The rapid development of Artificial Intelligence (AI) has brought about signifi ... Expand

Abstract: The rapid development of Artificial Intelligence (AI) has brought about significant changes in various industries. However, the opaqueness of AI systems leads to blind trust, distrust, or even avoidance of the systems altogether. Against this background, counterfactual explanation methods generate humanunderstandable explanations of how an AI system arrives at a given output. However, the thorough integration of categorical variables in the generation of counterfactual explanations remains a challenging research problem. In this paper, we investigate by means of a user study how a human-like handling of categorical variables affects users' perception of explanations for a state-of-theart counterfactual explanation approach. The results show that using a humanlike handling of categorical variables leads to the generation of explanations preferred by users. Moreover, we find that the handling of categorical variables affects users' perceptions of the concreteness, coherence, and relevance of the generated counterfactual explanations. Collapse

Topics: explainable artificial intelligence artificial intelligence electronic market

Methods: gradient boosting computational algorithm Student's t-test survey design artifact

How to Maximize Clicks for Display Advertisement in Digital Marketing? A Reinforcement Learning Approach

2023 | Information Systems Frontiers | Citations: 0

Authors: Singh, Vinay; Nanavati, Brijesh; Kar, Arpan Kumar; Gupta, Agam

Abstract: One of the core challenges in digital marketing is that the business conditions ... Expand

Abstract: One of the core challenges in digital marketing is that the business conditions continuously change, which impacts the reception of campaigns. A winning campaign strategy can become unfavored over time, while an old strategy can gain new traction. In data driven digital marketing and web analytics, A/B testing is the prevalent method of comparing digital campaigns, choosing the winning ad, and deciding targeting strategy. A/B testing is suitable when testing variations on similar solutions and having one or more metrics that are clear indicators of success or failure. However, when faced with a complex problem or working on future topics, A/B testing fails to deliver and achieving long-term impact from experimentation is demanding and resource intensive. This study proposes a reinforcement learning based model and demonstrates its application to digital marketing campaigns. We argue and validate with actual-world data that reinforcement learning can help overcome some of the critical challenges that A/B testing, and popular Machine Learning methods currently used in digital marketing campaigns face. We demonstrate the effectiveness of the proposed technique on real actual data for a digital marketing campaign collected from a firm. Collapse

Topics: marketing management artificial intelligence privacy big data web page

Methods: computational algorithm simulation experiment reinforcement learning descriptive statistic

Modeling relationships between retail prices and consumer reviews: A machine discovery approach and comprehensive evaluations

2021 | Decision Support Systems | Citations: 9

Authors: Yang, Xian; Yang, Guangfei; Wu, Jiangning; Dang, Yanzhong; Fan, Weiguo

Abstract: Setting the retail price as a part of marketing would affect customers’ cognitio ... Expand

Abstract: Setting the retail price as a part of marketing would affect customers’ cognition regarding products and affect their post-purchase behavior of review writing. To deeply understand the relationships between retail prices and reviews, this paper designs an intelligent data-driven Generate/Test Cycle using a machine learning technique to automatically discover the relationship model from a huge amount of data without a prior hypothesis. From a unique dataset, various free-form relationship models with their own structures and parameters have been discovered. By the comprehensive evaluations of candidate models, a guided map was offered to understand the relationship between dynamic retail prices and the volume/valence of reviews for different types of products. Experimental results show that 37.69% of products in our sample exhibit the following trend: When the price is increased to a certain level, the volume of reviews shifts from a decreasing trend to an increasing trend. Results also demonstrate that a linearly increasing relationship model between prices and the valence of reviews is more suitable for the low-involvement products than for the high-involvement products. In addition to the new findings, this research provides a powerful tool to assist domain experts in building relationship models for decision making in a highly efficient manner. Collapse

Topics: marketing management mobile phone website electronic commerce laptop computer

Methods: Monte Carlo simulation machine learning experiment mathematical model descriptive statistic

Hyperopic Search: Organizations Learning About Managers Learning About Strategies

2020 | Organization Science | Citations: 0

Authors: Ganz, Scott C.

Abstract: Early-stage experiments are central to the design-thinking approach to organizat ... Expand

Abstract: Early-stage experiments are central to the design-thinking approach to organizational innovation, and they are also a core practice in evidence-based management. Organizations use experiments to test new strategies in a low-stakes setting before escalating their commitment to a new initiative. Yet organizations also use experiments to evaluate managers who will implement these strategies in a high-stakes setting. I develop a formal model to demonstrate that these two types of evaluations are fundamentally incompatible. Managers who fear replacement in response to a poor experimental outcome pervert their experiments to maximize the likelihood that they succeed. This saps early experiments of much of their informational value. I show that if an organization can observe a manager's experimental strategies and the experiment’s outcomes and can commit ahead of time to an evaluation and replacement rubric, then it can resolve this agency problem. However, if a manager's experimental strategy cannot be credibly communicated, the organization will either replace good managers to induce a desirable experimental strategy or retain bad managers to alleviate the fear of replacement. I show that in low-uncertainty environments, the control exerted by the former replacement regime offers the best results, whereas, in high-uncertainty environments, commitment to retention regardless of experimental outcomes is best. Collapse

Topics: IT career task uncertainty organization learning interorganizational relationship organizational commitment

Methods: experiment triangulation genetic optimization optimization model stochastic model

Theories: organizational theory lemon market theory

A decision-making support system for Enterprise Architecture Modelling

2020 | Decision Support Systems | Citations: 0

Authors: Pérez-Castillo, Ricardo; Ruiz, Francisco; Piattini, Mario

Abstract: Companies are increasingly conscious of the importance of Enterprise Architectur ... Expand

Abstract: Companies are increasingly conscious of the importance of Enterprise Architecture (EA) to represent and manage IT and business in a holistic way. EA modelling has become decisive to achieve models that accurately represents behaviour and assets of companies and lead them to make appropriate business decisions. Although EA representations can be manually modelled by experts, automatic EA modelling methods have been proposed to deal with drawbacks of manual modelling, such as error-proneness, time-consumption, slow and poor re-adaptation, and cost. However, automatic modelling is not effective for the most abstract concepts in EA like strategy or motivational aspects. Thus, companies are demanding hybrid approaches that combines automatic with manual modelling. In this context there are no clear relationships between the input artefacts (and mining techniques) and the target EA viewpoints to be automatically modelled, as well as relationships between the experts' roles and the viewpoints to which they might contribute in manual modelling. Consequently, companies cannot make informed decisions regarding expert assignments in EA modelling projects, nor can they choose appropriate mining techniques and their respective input artefacts. This research proposes a decision support system whose core is a genetic algorithm. The proposal first establishes (based on a previous literature review) the mentioned missing relationships and EA model specifications. Such information is then employed using a genetic algorithm to decide about automatic, manual or hybrid modelling by selecting the most appropriate input artefacts, mining techniques and experts. The genetic algorithm has been optimized so that the system aids EA architects to maximize the accurateness and completeness of EA models while cost (derived from expert assignments and unnecessary automatic generations) are kept under control. Collapse

Topics: decision making decision support system database system source code anonymity

Methods: computational algorithm literature study descriptive statistic case study design science

Two-stage consumer credit risk modelling using heterogeneous ensemble learning

2019 | Decision Support Systems | Citations: 0

Authors: Papouskova, Monika; Hajek, Petr

Abstract: Modelling consumer credit risk is a crucial task for banks and non-bank financia ... Expand

Abstract: Modelling consumer credit risk is a crucial task for banks and non-bank financial institutions to support decision-making on granting loans. To model the overall credit risk of a consumer loan in terms of expected loss (EL), three key credit risk parameters must be estimated: probability of default (PD), loss given default (LGD) and exposure at default (EAD). Research to date has tended to model these parameters separately. Moreover, a neglected area in the field of LGD/EAD modelling is the application of ensemble learning, which by benefitting from diverse base learners reduces the over-fitting problem and enables modelling diverse risk profiles of defaulted loans. To overcome these problems, this paper proposes a two-stage credit risk model that integrates (1) class-imbalanced ensemble learning for predicting PD (credit scoring), and (2) an EAD prediction using a regression ensemble. Furthermore, multi-objective evolutionary feature selection is used to minimize both the misclassification cost (root mean squared error) of the PD and EAD models and the number of attributes necessary for modelling. For this task, we propose a misclassification cost metric suitable for consumer loans with fixed exposure because it combines opportunity cost and LGD. We show that the proposed credit risk model is not only more effective than single-stage credit risk models but also outperforms state-of-the-art methods used to model credit risk in terms of prediction and economic performance. Collapse

Topics: peer-to-peer model enterprise information system missing data

Methods: computational algorithm ensemble learning meta learning machine learning support vector machine

By the numbers: The magic of numerical intelligence in text analytic systems

2018 | Decision Support Systems | Citations: 3

Authors: Gruss, Richard; Abrahams, Alan S.; Fan, Weiguo; Wang, G. Alan

Abstract: There is a growing recognition among MIS researchers and practitioners that soci ... Expand

Abstract: There is a growing recognition among MIS researchers and practitioners that social media provide a valuable source of business intelligence. Unearthing relevant and useful information among the voluminous postings remains a challenge, however. Automated methods based on text mining have made significant progress in recent years by discovering a variety of new methods and features. This study adds to this stream by introducing a novel text mining procedure centered around numerical expressions contained in text documents. In this method, numerical expressions are extracted, categorized, and binned, and their presence and magnitude are stored as document features. We demonstrate, using a case study from the automotive industry, that numerical expressions can be reliably identified, and that these numerical features enable improvements in document classification. As an extension to this case study, we contribute a decision support system for managing product quality using both textual and numerical attributes. Collapse

Topics: social media decision support system information retrieval Python logistics management

Methods: case study machine learning logistic regression natural language processing experiment

A Literature Analysis of Research on Artificial Intelligence in Management Information System (MIS)

2018 | Americas Conference on Information Systems | Citations: 7

Authors: Nascimento, Alexandre Moreira

Abstract: This article presents an overview of Artificial Intelligence (AI) research appli ... Expand

Abstract: This article presents an overview of Artificial Intelligence (AI) research applied in the Management Information Systems field by analyzing the databases of the AIS journal basket list representing the major international journals on MIS. An initial set of 438 published papers was identified based on our search criteria. From those, 74 were selected because of their appropriate fit with the research questions. They were organized into domains, type of research, data collection strategy and AI technique. We found: (1) MIS Quarterly published 42% of the papers; (2) 50% of studies were on Information Systems applications, followed by 15% on Knowledge Management; and (3) 7 AI techniques were presented. This paper provides an overview of applied AI in Management, highlights the most studied related topics and techniques and contributes to a research agenda on the subject. Collapse

Topics: artificial intelligence marketing management fuzzy logic database system supply chain management

Methods: computational algorithm artificial neural network time series analysis longitudinal research literature sample

Theories: fuzzy set theory organizational behavior theory complex adaptive system

An evolutionary system for ozone concentration forecasting

2017 | Information Systems Frontiers | Citations: 5

Authors: Castelli, Mauro; Gonçalves, Ivo; Trujillo, Leonardo; Popovič, Aleš

Abstract: Nowadays, with more than 50 % of the world’s population living in urban areas, c ... Expand

Abstract: Nowadays, with more than 50 % of the world’s population living in urban areas, cities are facing important environmental challenges. Among them, air pollution has emerged as one of the most important concerns, taking into account the social costs related to the effect of polluted air. According to a report of the World Health Organization, approximately seven million people die each year from the effects of air pollution. Despite this fact, the same report suggests that cities could greatly improve their air quality through local measures by exploiting modern and efficient solutions for smart infrastructures. Ideally, this approach requires insights of how pollutant levels change over time in specific locations. To tackle this problem, we present an evolutionary system for the prediction of pollutants levels based on a recently proposed variant of genetic programming. This system is designed to predict the amount of ozone level, based on the concentration of other pollutants collected by sensors disposed in critical areas of a city. An analysis of data related to the region of Yuen Long (one of the most polluted areas of China), shows the suitability of the proposed system for addressing the problem at hand. In particular, the system is able to predict the ozone level with greater accuracy with respect to other techniques that are commonly used to tackle similar forecasting problems. Collapse

Topics: online learning system

Methods: genetic optimization machine learning support vector machine Bonferroni correction Kolmogorov-Smirnov test

Ontology-Based Mapping for Automated Document Management: A Concept-Based Technique for Word Mismatch and Ambiguity Problems in Document Clustering

2015 | ACM Transactions on Management Information Systems | Citations: 0

Authors: Lee, Yen-Hsien; Hu, Paul Jen-Hwa; Tu, Ching-Yi

Abstract: Document clustering is crucial to automated document management, especially for ... Expand

Abstract: Document clustering is crucial to automated document management, especially for the fast-growing volume of textual documents available digitally. Traditional lexicon-based approaches depend on document content analysis and measure overlap of the feature vectors representing different documents, which cannot effectively address word mismatch or ambiguity problems. Alternative query expansion and local context discovery approaches are developed but suffer from limited efficiency and effectiveness, because the large number of expanded terms create noise and increase the dimensionality and complexity of the overall feature space. Several techniques extend lexicon-based analysis by incorporating latent semantic indexing but produce less comprehensible clustering results and questionable performance. We instead propose a concept-based document representation and clustering (CDRC) technique and empirically examine its effectiveness using 433 articles concerning information systems and technology, randomly selected from a popular digital library. Our evaluation includes two widely used benchmark techniques and shows that CDRC outperforms them. Overall, our results reveal that clustering documents at an ontology-based, concept-based level is more effective than techniques using lexicon-based document features and can generate more comprehensible clustering results. Collapse

Topics: knowledge representation operating system Microsoft Windows Java information retrieval

Methods: ontological modelling cluster analysis experiment literature sample existing ontologies