مدل سازی رهبری - پروژه ارتباط عملکرد: عملکرد پایه ای شعاعی، روش گاوسی و کریگینگ به عنوان جایگزین هایی برای رگرسیون خطی
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|24631||2013||9 صفحه PDF||سفارش دهید||5464 کلمه|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Expert Systems with Applications, Volume 40, Issue 1, January 2013, Pages 272–280
The purpose of this paper is to analyze alternative forecasting methods that produce results at least similar to or better than linear regression (MLR) that can be used in the modeling of social systems. While organizations may be considered as typically non-linear systems, the common feature of most models found in literature continues to be the use of linear regression techniques. From a case study, advanced statistical methods of Gaussian and Kriging are evaluated, as well as an artificial intelligence (AI) tool, the radial basis function (RBF). The results show the best performance of the suggested methods compared to MLR, especially RBF, because of its uniform prediction behavior throughout all ranges of evaluation. These techniques, although somewhat unconventional in social systems modeling, present a potential contribution in increasing the accuracy and precision of the predictions allowing a more accurate assessment of the impact of certain strategies on the project performance to be made before the allocation of material, human and financial resources.
Much has been said about the need for understanding the influence of human contribution to the success of organizations. There is an increasing concern with the provision of services, the performance of people, and their effectiveness. Increasingly, the term “do more with fewer resources” gains strength. This approach has been proven to be valid in many situations but does not always yield the expected results. Many of these unsatisfactory results are only realized after the implementation of a project involving a large allocation of human and material resources and financial support (investments) for its execution. Large companies typically present corporate structures, and as such, they must be flexible. The corporation also has to be agile to cope with market competition, anticipate the competition, minimize the threats to its survival, and turn such threats into opportunities (Borgatti and Foster, 2003 and Vega and Vokurka, 2000). The interaction between leadership, agility and flexibility, and organizational factors contribute to these results, making a tool for predicting project performance essential to ensure the assertiveness of the process (Petri, 2004 and Singh et al., 2000). This environment has motivated researchers and managers to search for prediction models that allow the assessment of the impact of the people factor in organizations’ performances. Simulation models are powerful analysis tools for the development of research on the behavior of complex systems (Gilbert, 2007 and Gilbert and Troitzsch, 1999). However, the reliability of the predictions depends on the proper specification of the model variables and the choice of an appropriate modeling tool (Nikolopoulos & Assimakopoulos, 2003). While organizations may be considered to be typically non-linear systems, the basic feature common to most models in the social system field continues to be the use of traditional statistical techniques, such as multiple linear regression (MLR). This is due to its easiness to implement and broad dissemination of the field of social research and science management. It is also due to the mentioned fields’ scholars and their lack of understanding of the use of more sophisticated alternative methods (Harrison et al., 2007 and Somers and Casal, 2009). Nevertheless, the introduction of these analysis methods, whether they are more elaborate statistical methods or even artificial intelligence tools, aims to overcome the limitations found in conventional statistical techniques. These restrictions are associated with the assumption that there is a linear relationship amongst the system variables - in other words, the effect of a dependent variable is proportional to the sum of a set of independent variables (Gilbert and Terna, 2000 and Terziovski, 2002). The purpose of this paper is to analyze alternative forecasting methods that produce results at least similar to or better than linear regression methods, and that can be more frequently used in the modeling of social systems. Three groups of tools are evaluated and compared so that, by the end of this study, it is possible to describe the impact of the used technique in the accuracy of the results obtained through simulation. The first group concerns with the traditional regression techniques, while the others are related to the introduction of advanced statistical methods and artificial intelligence (AI) tools. Section 2, which follows after this introduction, describes the proposed method, where the variables of the research are presented, the modeling tools used, and the method of analysis. Section 3 deals with the empirical study, starting with certain considerations about the sample used and a discussion of the results. The conclusions of the study are found in Section 4.
نتیجه گیری انگلیسی
The results of this study help to clarify the idea of the viability of the usage of advanced statistical methods, such as Kriging and Gaussian process, and artificial intelligence tools (AI), like the RBF, for modeling in science management. Although such tools are somewhat unconventional to be used in modeling social systems, there is a potential contribution in increasing the accuracy and precision of the predictions that can be obtained by applying the models, even when considering the complexity and the nonlinearity of these systems. Another important conclusion is that none of the proposed methods presented the ability to predict results inferior to that of the linear regression method (MLR). On the contrary, such methods were shown to be either better than or at least equivalent to the MLR. This confirms the expectations presented earlier in this article. It is suggested that the choice of a forecasting tool to begin with an analysis of the use of linear regression techniques due to its wide application range, simplicity, and good predictive capability in systems with low nonlinearity. Since the results obtained from linear regression do not meet the levels of accuracy and precision required for the process analysis (high non-linearity), the use of alternative methods such as those mentioned in this study is proposed. It is believed that the impact of the presented results is relevant from both the academic and organizations point of views. From the academic perspective, contribution is expected to encourage a more frequent use of alternative modeling techniques with higher accuracy and precision of predicting the outcomes, especially for highly non-linear systems. It is also expected a contribution on a new evaluation of the consideration that traditional statistical tools (such as linear regression and correlation analysis) are always enough appropriate for the modeling of social systems. From the perspective of the organizations, more sophisticated modeling allow a more accurate assessment of the impact of certain strategies (scenarios) on the performance of projects, to be made before the actual allocation of material and human resources, consequently enabling a better profile match of manpower, investments and projects.