Are Evaluation of ICTs Investment Using Decision Support Systems:Case Applications from Distributor's and End User's Perspective Group Decision

The evaluation of investments in advanced technology is one of the most important decision making tasks. The importance is even more pronounced considering the huge budget concerning the strategic, economic and analytic justification in order to shorten design and development time. Choosing the most appropriate technology requires an accurate and reliable system that can lead the decision makers to obtain such a complicated task. Currently, several Information and Communication Technologies (ICTs) manufacturers that design global products are seeking local firms to act as their sales and services representatives (called distributors) to the end user. At the same time, the end user or customer is also searching for the best possible deal for their investment in ICT's projects. Therefore, the objective of this research is to present a holistic decision support system to assist the decision maker in Small and Medium Enterprises (SMEs) - working either as individual decision makers or in a group - in the evaluation of the investment to become an ICT's distributor or an ICT's end user. The model is composed of the Delphi/MAH (Maximising Agreement Heuristic) Analysis, a well-known quantitative method in Group Support System (GSS), which is applied to gather the average ranking data from amongst Decision Makers (DMs). After that the Analytic Network Process (ANP) analysis is brought in to analyse holistically: it performs quantitative and qualitative analysis simultaneously. The illustrative data are obtained from industrial entrepreneurs by using the Group Support System (GSS) laboratory facilities at Lappeenranta University of Technology, Finland and in Thailand. The result of the research, which is currently implemented in Thailand, can provide benefits to the industry in the evaluation of becoming an ICT's distributor or an ICT's end user, particularly in the assessment of the Enterprise Resource Planning (ERP) programme. After the model is put to test with an in-depth collaboration with industrial entrepreneurs in Finland and Thailand, the sensitivity analysis is also performed to validate the robustness of the model. The contribution of this research is in developing a new approach and the Delphi/MAH software to obtain an analysis of the value of becoming an ERP distributor or end user that is flexible and applicable to entrepreneurs, who are looking for the most appropriate investment to become an ERP distributor or end user. The main advantage of this research over others is that the model can deliver the value of becoming an ERP distributor or end user in a single number which makes it easier for DMs to choose the most appropriate ERP vendor. The associated advantage is that the model can include qualitative data as well as quantitative data, as the results from using quantitative data alone can be misleading and inadequate. There is a need to utilise quantitative and qualitative analysis together, as can be seen from the case studies.

Kernel-Based Ranking. Methods for Learning and Performance Estimation

Machine learning provides tools for automated construction of predictive models in data intensive areas of engineering and science. The family of regularized kernel methods have in the recent years become one of the mainstream approaches to machine learning, due to a number of advantages the methods share. The approach provides theoretically well-founded solutions to the problems of under- and overfitting, allows learning from structured data, and has been empirically demonstrated to yield high predictive performance on a wide range of application domains. Historically, the problems of classification and regression have gained the majority of attention in the field. In this thesis we focus on another type of learning problem, that of learning to rank. In learning to rank, the aim is from a set of past observations to learn a ranking function that can order new objects according to how well they match some underlying criterion of goodness. As an important special case of the setting, we can recover the bipartite ranking problem, corresponding to maximizing the area under the ROC curve (AUC) in binary classification. Ranking applications appear in a large variety of settings, examples encountered in this thesis include document retrieval in web search, recommender systems, information extraction and automated parsing of natural language. We consider the pairwise approach to learning to rank, where ranking models are learned by minimizing the expected probability of ranking any two randomly drawn test examples incorrectly. The development of computationally efficient kernel methods, based on this approach, has in the past proven to be challenging. Moreover, it is not clear what techniques for estimating the predictive performance of learned models are the most reliable in the ranking setting, and how the techniques can be implemented efficiently. The contributions of this thesis are as follows. First, we develop RankRLS, a computationally efficient kernel method for learning to rank, that is based on minimizing a regularized pairwise least-squares loss. In addition to training methods, we introduce a variety of algorithms for tasks such as model selection, multi-output learning, and cross-validation, based on computational shortcuts from matrix algebra. Second, we improve the fastest known training method for the linear version of the RankSVM algorithm, which is one of the most well established methods for learning to rank. Third, we study the combination of the empirical kernel map and reduced set approximation, which allows the large-scale training of kernel machines using linear solvers, and propose computationally efficient solutions to cross-validation when using the approach. Next, we explore the problem of reliable cross-validation when using AUC as a performance criterion, through an extensive simulation study. We demonstrate that the proposed leave-pair-out cross-validation approach leads to more reliable performance estimation than commonly used alternative approaches. Finally, we present a case study on applying machine learning to information extraction from biomedical literature, which combines several of the approaches considered in the thesis. The thesis is divided into two parts. Part I provides the background for the research work and summarizes the most central results, Part II consists of the five original research articles that are the main contribution of this thesis.