一般化巢式羅機模式參數眾多，模式結構複雜，一般非線性求解法難以順利校估求得模式的巢層架構及全部的參數值，故只能嘗試以其他演算法來校估模式。本研究嘗試設計啟發式求解法與基因演算法來校估一般化巢式羅機模式，比較兩種演算法的效率與校估結果的正確性；並以國道客運台北--高雄線的旅運選擇資料為實證資料，利用一般化巢式羅機模式分析各家客運業者間的相關性。

本研究首先設計適合用於校估一般化巢式羅機模式的基因演算法演算機制，找出最有效率的演算流程與演算方式，以基因演算法找到合適的起始解後，再代入非線性規劃法中校估求得各參數的精確值及統計指標。提出啟發式求解法的概念及求解方式，將全部巢層分成兩類後再分別處理，以減少一次所需校估的巢層數。利用兩種方法校估一般化巢式羅機模式，比較兩種演算法的求解結果，最後得到結論如下：

1. 啟發式求解法與基因演算法都能順利校估出一般化巢式羅機模式的參數值，兩種演算法最後所得到的模式之巢層架構相同，各個參數值也都一樣，表示兩種演算法皆正確且可行。
2. 由於兩種演算法皆可行，研究者在選擇採用何種演算法時，需審慎考慮需求及可供利用之電腦設備做最適當的選擇。整體而言，採用啟發式求解法所需花費的校估時間較少，且只需一台電腦即可進行，但校估過程較繁瑣。採用基因演算法時，最好能有多台電腦可同時進行校估，雖然所需花費的校估時間較長，但較容易進行。

The model structure of the generalized nested logit model is very complex. It is very difficult to calibrate its parameters using general non-linear estimation methods. This research designed two calibrating methods to tackle this problem, i.e., heuristic method and Genetic Algorithms approach. Their efficiency and outcomes are compared. We used bus travelers traveling between Taipei and Kaohsiung to analyze the unobserved correlations among bus companies serving this route.

First, we designed the algorithms which are suitable for the calibration of the generalized nested logit model and found the most efficient flow path and mode. Genetic Algorithms were used to search out the suitable initial solution for the non-linear estimation methods to get the final parameters. The idea of heuristic method is separating all nests into two parts and then dealing with them. The advantage of this method is its ability to reduce the number of nests to be estimated at each time. Comparing the results of heuristic method and Genetic Algorithm method, we found the following conclusions:

1. Heuristic method and Genetic Algorithm method got the same parameter estimates for the generalized nested logit model. This result showed that both methods are feasible and correct.

2. The choice of these two methods is dependent on the computer resources. The heuristic method needs only one computer but the estimating procedure is very troublesome. Genetic Algorithm method is easier to proceed but needs more computers.

1. 段良雄(1984)，「巢式多項羅機(NMNL)運具選擇模式」，運輸計畫季刊，第十三卷，第三期，頁285-308。
2. 施鴻志、段良雄、凌瑞賢(1984)，「都市交通計畫—理論、實務」，茂昌圖書有限公司。
3. 周榮昌、林明德(1996)，「以基因演算法求解多項式普羅比模式之研究」，中華民國運輸學會第11屆學術研討會。
4. 溫傑華、藍武王、張仲杰(1999)，「以成對組合羅吉特模式探討城際間運具選擇行為」，中華民國運輸學會第14屆學術研討會。
5. 溫傑華(2000)，「以一般化巢式羅吉特模式探討城際間運具選擇之研究」，中華民國運輸學會第15屆學術研討會。
6. 方菀萍(2002)，「個體選擇模式選擇集合之研究—以國道客運北高路線為例」，國立成功大學交通管理科學研究所碩士論文。
7. 曾維琦(2002)，「國道客運公司市場定位之研究—以北高線為例」，國立成功大學交通管理科學研究所碩士論文。
8. 謝欣宏(2002)，「台鐵司機員排班與輪班問題之研究—以基因演算法求解」，國立成功大學交通管理科學研究所碩士論文。
9. Bhat, C.R. (1995), “A heteroscedastic extreme value model of intercity mode choice”, Transportation Research B, 29, pp.471-483.
10. Michalewicz, Z. (1996), “Genetic Algorithms + Data Structures = Evolution Programs”, 3rd Revised and Extended Edition., Springer-Verlag, 1996.
11. Vovsha, P. (1997), “The cross-nested logit model: application to mode choice in the Tel-Aviv metropolitan area”, Transportation Research Record 1607, pp.6-15.
12. Koppelman, F.S. & Wen C. (1998), “Alternative nested logit models: structure, properties and estimation”, Transportation Research B, 32, pp.289-298.
13. Koppelman, F.S. & Wen C. (1999), “Nested logit models: which are you using？”, Transportation Research Record 1645, pp.1-7.
14. Koppelman, F.S. & Wen C. (2000), “The paired combinatorial logit model: properties, estimation and application”, Transportation Research B, 34, pp.75-89.
15. Liu, Y. & Mahmassani, H.S. (2000), “Global maximum likelihood estimation procedure for multinomial probit (MNP) model parameters”, Transportation Research B, 34, pp.419-449.
16. Koppelman, F.S. & Wen C. (2001), “The generalized nested logit model”, Transportation Research B, 35, pp.627-641.
17. Ortuzar, J. de D. (2001), “On the development of the nested logit model”, Transportation Research B, 35, pp.213-216.
18. Bhat, C.R.(2001), “Quasi-random maximum simulated likelihood estimation of the mixed multinomial logit model”, Transportation Research B, 35, pp.677-693.
19. Carrasco, J.A. & Ortuzar, J. de D. (2002), “Review and assessment of the nested logit model”, Transport Review, 22(2), pp.197-21.