بهینه سازی تکاملی جستجوی مشخصه های مدل بین دانش مدیریت و عملکرد مهندسی ساخت
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی|
|3333||2013||13 صفحه PDF||30 صفحه WORD|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Expert Systems with Applications, Volume 40, Issue 11, 1 September 2013, Pages 4414–4426
2. مروری بر مقالات
3. پیش فرض های پژوهش
4.1 مدل سازی معادلات ساختاری
4.2. قابلیت اطمینان و تحلیل تأییدی
شکل 1: مدل فرضی
4.3 اعمال GA در جستجوی مشخصه مدل
5. فرآیند تحلیلی و نتایج
5.1 آماره های توصیفی
جدول 1. اصطلاحات GA-SEM
جدول 2 مشخصه های اجتماعی- اقتصادی پاسخ دهندگان تحقیق
5.2. تحلیل تجربی
جدول 3. تحلیل تجربی متغیرها
شکل 3. کاربرد GA در SEM
5.3 بهینه سازی جستجوی مشخصه مدل
5.3.1 معیاری برای برازندگی قضاوت
184.108.40.206 مربع کی/ درجات نسبت آزادی
220.127.116.11. شاخص نیکویی برازش (GFI).
18.104.22.168. شاخص برازش افزایشی (IFI).
جدول 4. جدول کدگذاری گروه اول
22.214.171.124. شاخص برازش تطبیقی (GFI).
126.96.36.199. میانگین مربعات خطای برآورد (RMSEA
جدول 5 شاخص های GOF برای کروموزوم ها
شکل 4: نمودارهای همگرایی شاخص
جدول 6 نتایج تست مسیر برای فرضیه های پژوهش مربوطه
5.3.2. رمزگذاری و فرآیند تکامل
5.4. اصلاح مدل
شکل 5. مدل اصلاح شده
جدول 7. معیار شاخص GOF و مقادیر حاصل
جدول 8. تأثیر روش های مدیریت بر روی عملکرد پروژه
6. نتیجه گیری و کارهای آتی
Many studies have attempted to specify alternative model configurations as fitting empirical data with the aid of structural equation modeling (SEM) method. However, significant path searching between constructs has increased in difficulty and complexity. One way to enhance modeling efficiency is evolutionary optimization by genetic algorithm (GA). This study applies the project management (PM) knowledge possessed by construction personnel and uses techniques, tools, and skills (TTS) to explore the causal relationship between TTS usage and construction engineering project performance (PP). A questionnaire survey is used to empirically measure the effectiveness of PM TTS on PP. The research framework is first defined by hypotheses supported by the literature. The GA is then applied to the model fitting process to optimize the structural paths. Analytical results show that evolutionary optimization for singular and multiple goodness of fit effectively searches the SEM specifications. By using GA in SEM procedure, researchers can perform automated specification searches to find the best empirical model fit to the data.
In the behavioral social science domain, numerous hypothetical constructs cannot be measured or known by observation alone. For instance, quality, communication, risk, satisfaction, and success are intangible characteristics or abstract constructs. These constructs can only be observed indirectly by measuring indicators that reflect the characteristics of the constructs. Thus, structural equation modeling (SEM) was created to evaluate and analyze causal relationships between latent constructs and measurable indicators. The SEM technique, which originates from confirmatory factor analysis and structural path analysis, was initially proposed by Joreskog (1973) and is now a widely used research tool in psychology (Anderson, Babin, Black, & Hair, 2010), social sciences (Fitch, 2007), health sciences (Gonzalez-de la Parra, Namur, & Rodriguez-Loaiza, 2006), and management (Chinda and Mohamed, 2008 and Hsu and Sabherwal, 2011). However, the use of SEM to analyze the path between different constructs must consider both direct and indirect effects. Thus, the difficulty of the search for specific structures increases as the number of path alternatives in the hypothetical research model increases. Traditionally, model specifications are searched manually, which is time-consuming and inefficient. Automating the search process facilitates management of this chaotic procedure. For the above reasons, this study applied genetic algorithm (GA), an adaptive heuristic search procedure for processing large-scale optimization problems. Model specifications are optimized in a case study of causal linkages between project management (PM), techniques/tools/skills (TTS), PM knowledge and construction engineering project performance (PP). Additionally, although the PM approach includes the fundamental practices needed for success in the construction industry, it does not consider the effectiveness of PM TTS that are considered useful to practitioners in terms of project performance. This study fills this gap in the literature. The four research stages were research assumptions, questionnaire design and analysis, structure optimization, model modification and discussion. Detailed descriptions are as follows: Step 1 – Research assumptions: set research objectives and explore the relationship between PM TTS, PM knowledge, and PP through a literature review followed by construction of the research structure model. Step 2 – Questionnaire design and analysis: distribute design questionnaires to interviewees. After retrieving questionnaires, perform statistical analyses, including descriptive statistical analysis, reliability analysis, validity analysis, confirmatory analysis, path analysis and normality testing. Step 3 – Optimize the overall structural model: use GA-SEM to determine the optimal structure specifications. This study applies GA to improve the initially assumed structural path by executing evolutionary procedures. Step 4 – Model modification and discussion: after completing verification in the previous step, modify the optimized structure model to fit the empirical data. Finally, the effects of PM TTS on project performance are analyzed and discussed.
نتیجه گیری انگلیسی
This study integrates evolutionary genetic algorithm with structural equation modeling (SEM) to perform model specification search optimization. Previous studies used time-consuming manual search methods to explore the model structure. The research method proposed in this study is based on adaptive heuristic search procedures, which can process large-scale optimization problems. Using automated procedures and algorithms facilitates the search for the optimal structure path. Surprisingly, the analytical results of the demonstrated model show that not all the project management body of knowledge (PMBOK) techniques/tools/skills (TTS) have significant stacking effect on construction engineering project performance (PP) but it does have mutual inter-correlations between some PMBOK. For PP, only project communications management (PCoM) and project procurement management (PPM) have direct and statistically significant impacts according to the analytical results. Confirmatory analysis in the final model showed that, PCoM can be assessed empirically and effectively by “stakeholder analysis”, “communication requirements” and “communication methods”. Likewise, PPM can be accessed by “bidder conference” and “procurement negotiations”. Thus, project managers attempting to improve PP should use these TTS preferably if available resources are limited. This study makes two contributions to the domain knowledge. For researchers using SEM, the methods proposed in this paper enable efficient determination of structural model paths. For the construction industry, the findings of this investigation enable experts to improve PP, increase company owner satisfaction, and facilitate effective use of management resources under limited capital, time, and other relevant resources by industry workers, all of which enhance the efficiency of resource use. For future directions, periodical surveys are suggested for monitoring social and technological changes in management techniques, tools, and skills. Similarly, future work can adopt the proposed systematic approach in cross country comparisons or in engineering applications. A further structural model linking client satisfaction and project success as consequences is another intriguing research theme in construction engineering management.