5 results
Search Results
2. Modeling Decisions for Artificial Intelligence : 12th International Conference, MDAI 2015, Skövde, Sweden, September 21-23, 2015, Proceedings.
- Author
-
Narukawa, Yasuo and Torra, Vicenc
- Subjects
Application software ,Artificial intelligence ,Data mining ,Information storage and retrieval ,Numerical analysis ,Pattern recognition - Abstract
Summary: This book constitutes the proceedings of the 12th International Conference on Modeling Decisions for Artificial Intelligence, MDAI 2015, held in Skövde, Sweden, in September 2015. The 18 revised full papers presented were carefully reviewed and selected from 38 submissions. They discuss theory and tools for modeling decisions, as well as applications that encompass decision making processes and information fusion techniques.
- Published
- 2015
3. Exploring the relationship between the Social Network Profile of S&P1500 Firms and their environmental and financial profiles
- Author
-
Shahgholian, Azar, Theodoulidis, Charalampos, and Papamichail, Konstantinia
- Subjects
Social Network of S&P 1500 companies ,Data Mining ,Text Mining ,Financial Profile ,Environmental Profile - Abstract
The purpose of this thesis is to enhance our understanding of the relationship between the social network profile of S&P1500 firms and their environmental and financial profiles. The three dimensions of this research are social network profile, financial profile and environmental profile, which are becoming increasingly interlinked. The nature of this research is multidisciplinary and still in its early stages. The existing studies focus mainly on two streams of research, the first of which explores the relationship between firms' environmental and financial profiles, which reveals contradictory results. The second research stream investigates the impact of the social networks between directors on the firms' financial profiles. This thesis is submitted in an alternative format and includes four journal papers, which are interrelated in addressing the purpose of this thesis. First, it is essential to provide effective reviews to create a foundation for developing knowledge in this field of research and to explore the area in which more research is required. Therefore, the first two papers attempt to review systematically the existing research streams, namely: (i) the impact of social network profile on financial profile; and (ii) the relationship between environmental profile and financial profile. Second, the review of the impact of social network profile on financial profile reveals the need to investigate social networks in the organisations from a social network theory perspective. Therefore, the third paper uses quantitative method to provide a concrete definition of social networks and social network centrality metrics in the context of organisations. In addition, a clear process for extracting social networks at both director and board levels from the directors' information repository is defined. Third, through the fourth paper, this thesis explores the impact of the board's roles on environmental governance as an essential component of environmental profile. This paper uses a combination of qualitative and quantitative evaluation. The roles of the board of directors in relation to environmental profile are twofold, namely board monitoring and board resource provision. In this work, the board's social network is examined as a board resource-provision role.
- Published
- 2016
4. Data Analytic Approaches to Predicting Success in Bank Telemarketing
- Author
-
Lentz, Curtis L. and Lentz, Curtis L.
- Subjects
- Data mining., Telemarketing., Banks and banking., Data Mining, Exploration de données (Informatique), Télémarketing., Banks and banking., Data mining., Telemarketing.
- Abstract
This paper uses a large data set from a Portuguese bank to use a data analytic approach to improving the results of telemarketing campaigns. The data was collected from 2008 to 2013, and was obtained from the University of California at Irvine's Machine Learning Repository. It contains 41,118 records and 20 variables, including the target variable of "yes" as a customer's response to the banks offer of a long-term deposit contract. The modeling is performed using IBM's Modeler software. The first section of the paper establishes a foundation for Customer Relationship Management as a powerful and growing method of using data analysis techniques to better understand a company's customers and their relationship with the business. It also outlines the CRISP-DM approach to organizing a data mining project. The data is analyzed in detail, including performing Exploratory Data Analysis (EDA) on all variables to get a detailed view of the data and the distributions of the variables, as well as an understanding of the relationship between the variables and the dependent variable. Further, Principal Component Analysis and K-means clustering is conducted on the data to identify potential correlations in variables and how groups of customers might respond similarly. Logistical regression is used on four different models, the results are described and compared across a variety of criteria. Using the profit criteria, the model with the highest revenue per customer is selected as a model to be used for prediction of how to select the best potential customers for a future campaign. Four other modeling techniques: CART, C5.0, Support Vector Machines, and Neural Networks are tested to see if the predictive results can be improved. Logistic regression is recommended as the best approach due to its higher profit per customer, and its relative ease in interpreting its results over the other modeling techniques. The thesis ends with conclusions, recommendations and suggestions for future research.
- Published
- 2015
5. Essays on residential electricity consumption profiles : weather effects and household behaviour patterns
- Author
-
Kang, Jieyi and Reiner, David
- Subjects
333.79 ,smart metering ,data mining ,residential electricity ,consumption behaviour ,weather response - Abstract
The high temporal resolution data created by smart metering, which has now been deployed in many countries, provides an unprecedented opportunity to examine household consumption behaviour in narrow time windows, whereas past studies could only look at monthly or even yearly consumption. However, most studies that have used smart meter data focused either on load management (load forecasting, theft detection, etc.) or linked electricity usage to demographic and/or building characteristics. Few studies have been conducted on the impacts of weather on intraday consumption behaviour. Better appreciation of the influence of weather could improve pricing designs as well as provide better understanding of household behaviour, which could, for example, potentially increase energy efficiency. With knowledge of weather effects on residential consumption, it could also be valuable for utilities to improve grid stability and reduce operation cost. To fill the gap, this dissertation analyses the impact of different weather variables as well as consumption patterns through different tools based on smart metering data. This thesis uses a three article format. Chapter 1 provides a general overview of the literature on smart meters and empirical studies using smart metering data. Chapter 2 presents an econometric analysis of the effect of weather factors in Ireland (such temperature, rainfall and sun duration) at different periods of a day, and contrasts the impacts on consumption for workdays versus weekends versus holidays. Chapter 3 employs machine learning methods - clustering algorithms - to categorise households by their electricity demand response to different weather variables. The results demonstrated that some weather sensitivity patterns are closely associated with household characteristics. In Chapter 4, smart meter data was gathered from a very different location, Chengdu, the capital of Sichuan Province in China, which has more extreme weather and greater variability. Three scenarios are analysed in Chapter 4: (1) weekly consumption profiles in different seasons; (2) festival (major holiday) consumption profiles; and (3) consumption patterns during extreme weather. Finally, the thesis is concluded by Chapter 5, which summarises the main empirical and methodological contributions of the three papers and lays out future work in this area.
- Published
- 2020
- Full Text
- View/download PDF
Discovery Service for Jio Institute Digital Library
For full access to our library's resources, please sign in.