從民國87年所通過的強制汽車責任保險，可發現我國強制汽車責任險的理賠上限與承保範圍等有了重大的改變，造成汽車任意責任險的計算費率必須要重新估算，然而國內對此方面的文獻探討卻著墨不多，因此學生將會針對任意汽車體傷責任險費率釐算詳細加以探討。 而若要重新估計任意汽車責任險首先要做的工作就是要收集完整正確的損失資料，不過由於損失資料的收集相當困難，因此只能透過模擬的損失資料進行任意責任險的費率釐算。而在有模擬的損失資料情況下我們就可透過損失分佈理論進行下列的分析： (1)透過損失資料的特性推估任意汽車責任險可能之損失分佈為Lognormal 分佈。 (2)透過最大概似估計法與特殊法推估Lognormal分佈之參數，並且採用負對數蓋似函數選擇最佳之估計參數。 (3)透過與強制汽車責任險預期損失與汽車任意體傷責任險預期損失之比例關係，釐算汽車任意體傷責任險之純保費。 (4)建立兩種損失趨勢函數，並透過此兩趨勢函數計算汽車任意體傷責任險之高保額係數。 (5)透過上述步驟之計算結果與現行實施之汽車任意體傷責任險費率作比較，以探討是否現行費率是否有超收或不足的現象。 總之，希望此篇論文能夠對未來的汽車任意責任險之費率釐算與保險司費率監督有所幫助。 / Cause the Legislation Yuan passed the compulsory auto liability insurance bill in 1998, we must have a new actuarial pricing of voluntary auto third party insurance. However, all domestic insurers haven’t revised the rate because the absence of the empirical loss data. In addition, only a fewer researches have focused on the actuarial model of this type of insurance. In this paper, we will investigate the pure premium calculation of the voluntary auto insurance, and outline the appropriate model construction procedures. The data we use are not empirical loss data, we calculate the pure premium by the simulated data. The procedures of this study are summarized in the following: (1) Find the possible loss distribution of voluntary auto third party insurance policy. (2) Estimate the parameters of the loss distribution by the maximum likelihood estimate method and the special method of lognormal distribution. (3) Calculation the pure premium of voluntary auto third party insurance. (4) Calculation the increased limits factor(ILF)by two trend functions, and compare the results of two trend functions. (5) Finally, we examine the gross premiums of the voluntary auto third party insurance and compare our results with the actual voluntary auto liability insurance premiums. Altogether, we hope that this paper could be beneficial to the actuaries and also provide suggestions for the government surveillance.

在醫學領域中，磁振造影（Magnetic Resonance Imaging, MRI）因為具有良好的空間解析度及對比度，且不會對人體產生任何輻射性或侵入性的傷害，所以在疾病診斷中為經常被醫師們使用的輔助工具。其中利用磁振造影測量患者腦部血流情形所攝得之對比劑濃度與時間關係曲線圖，更是醫學界在對付腦血管病變（Brain Lesion）時的診斷利器。然而截至目前為止，我們尚未有一個較正確且快速的方法可以用來配適其對比劑濃度與時間關係曲線中的參數。所以在本論文中，我們嘗試以統計上的觀點，利用幾種不同的配適方法，找出與原始觀察值最為接近之估計值。 在本研究中使用的配適方法有—「迴歸分析法」、「Whittaker修勻法」、「非線性函數參數修勻法」及「核修勻法（Kernel Graduation）」。 本論文將以往醫學界慣用的「乘方性誤差項」改變為「加成性誤差項」，再以不同的誤差項，利用電腦模擬出各組假資料（Pseudo Data）後，以上述的四種方式對原始觀察值進行參數配適與函數估計。綜合模擬資料與真實資料所配適的比較結果，我們認為在幾種方法中，最穩健（Robust）的配適法是「Whittaker修勻法」。而在本論文中進行配適的真實資料，應該具有較大的誤差項，才導致非線性函數參數修勻法不能得出很好的估計值。 / With greater resolution, higher contrast and no radiative hurt to human body, Magnetic Resonance Imaging (MRI) is widely used by doctors in diagnosing diseases. The concentration of the contrast agent v.s. time curves which generated by MRI for cerebral blood flowing is very useful to doctors when giving treatments to brain lesion. However, we still have no precise and quick solution for fitting the curve of the concentration of the contrast agent vs. time. Therefore, this essay tries to use some different statistical fitting methods to find the closest estimates to the crude observations. We will use four different fitting methods here—"Regression Analysis", "Whittaker Graduation", "Nonlinear Function Parametric Graduation", "Kernel Graduation". This essaywill change the "multiple error term" which was usually used in the medical field to "additive error term". After using different sizes of error terms to generate pseudo data by computer simulation, we fit the parameters and estimate the values of the function to the crude data we've created with the four fitting methods mentioned above. Comparing the fitting result of the simulation data and the real data, we think the most robust fitting method is " Whittaker Graduation". The real data we have fitted in this essay may contain a greater error term, it would make " Nonlinear Function Parametric Graduation" get inadequate fitting values.

犯罪行為受到警力或法律執行的影響，會移動到鄰近警力較差地區。正如 Gylys所說：考量一個地區警力的多寡，將受到其他鄰近區域警力的影響 很大。Mehay亦認為：從實際經驗上來看，對於移動性的犯罪(如搶劫、縱 火、偷竊等)，外部支配型式力量(如警力)的適當增加，將迫使其外溢( spillovers)至鄰近區域。利用空間統計的自我迴歸模式，我們可以更了 解移動性犯罪受到相連區域自我相關的影響。即使相關性不高，在作了差 分之後，其主成分分析最大負載變數項，變化相當大。所以資料裡，如果 有區域自我相關的情形，就應該謹慎處裡。

無人飛行載具(Unmanned Aerial Vehicle, UAV)於要求精度之圖資測製應用時，因飛行高度較低並可在雲下飛行取像，與大型載具相比可更機動性獲取空間解析度較高之影像，雖無法如大型載具酬載大像幅感測器供大區域圖資製作，但於小區域之圖資更新卻相當適合。但一般UAV因酬載重量限制，僅可酬載體積小且重量輕之感測器，如非量測型相機及低精度定位定向系統，即AHRS系統。因此，本研究嘗試在UAV上酬載Trimble BD970 GNSS OEM GPS接收模組，此GPS接收模組體積小且重量輕可安置於UAV上，並透過VBS-RTK GPS定位技術獲取UAV精確飛行軌跡資訊，再經時間內插相機曝光瞬時的GPS資訊供空中控制使用，輔助UAV影像空中三角測量(簡稱空三)平差，以降低地面控制點需求。 但欲引入GPS觀測量供空中控制使用必須考量GPS天線與相機投影中心偏移量之問題，但因UAV所酬載之非量測型相機，將造成此偏移量不易透過地面測量方式測得，於本研究將於空三平差時使用線性漂移參數克服此偏移量無法量測之問題；此外，UAV所酬載之非量測型相機，相機參數乃透過地面近景攝影測量以自率光束法平差方式率定所得，但率定所得相機參數無法完全描述相機在航拍取像時的情況，故本研究於空三平差將採用自率光束法克服相機參數率定不完全之問題。實驗中，首先確定GPS模組BD970在VBS-RTK GPS定位技術下在地面高速移動時可獲得高精度的定位成果；接續驗證線性漂移參數及自率光束法平差於此研究的適用性；最後亦探討不同地面控制點配置及來源對空三平差之精度探討，並提出1/5000基本圖圖資測製精度要求下，VBS-RTK GPS輔助UAV影像自率光束法空三平差的地面控制點最適配置。 / UAV(Unmanned Aerial Vehicle) is currently used in civil purpose such as mapping and disaster monitoring. One of UAV advantages is to collect images with high resolution for mapping demand. However, due to payload limitations of UAV, it is difficult to mount metric aerial camera and precise POS(Positioning and Orientation System) device. Instead, only the non-metric camera and the low accurate AHRS (Attitude and Heading Reference System) can be installed. For mapping demands, Trimble BD970 GNSS OEM board will be carried on the UAV to collect the high accurate flying trajectory as control information for AT (aerial triangulation) by VBS-RTK(Virtual Base Station - Real Time Kinematic) GPS technique. Meanwhile self-calibration bundle adjustment will be employed for AT(Aerial Triangulation) to overcome the imperfect calibration of non-metric camera by the close-range photogrammetric approach. The precise offset between image perspective center and GPS antenna center, called GPS antenna-camera offset, is hard to measure in centimenter level by terrestrial measurement approach. Therefore the drift parameters will be utilized to solve the problem of GPS antenna-camera offset while performing bundle adjustment with self-calibration for AT of UAV images. In the experiments of this study, the height positioning accuracy of BD970 by VBS-RTK GPS approach at high speed movement will be proved firstly. Then the adaptability of drift parameters and self-calibration for GPS supported AT of UAV images will be verified. Finally, the accuracy of AT by using different control information will be analized and appropriate configuration of GCPs(Ground Control Points) for VBS-RTK GPS supported self-calibration bundle adjustment for AT of UAV images will be proposed under the mapping demand with the scale of 1 : 5000.

競爭風險之下對邊際分配的估計，是許多領域中常遇到的問題。由於主要事件及次要事件互相競爭，只要一種事件先發生即終止對另一事件的觀察，在兩事件同時發生的機率為0之下，連一筆完整的資料我們都無法蒐集到。除非兩事件互為獨立或加上其它條件，否則會有邊際分配無法識別的問題。但是獨立的條件在有些情況下並不合理，為解決相依競爭風險之邊際分配無法識別的問題，可先假定兩事件發生時間之間的關係。 由於關聯結構定義出兩變數間的結合關係，我們可利用關聯結構解釋兩事件發生時間之間的關係。假定兩變數之相關性參數為已知，且採用機率積分轉換的觀念，本論文討論了Zheng 與 Klein提出的關聯結構-圖形估計量，是否會依設限程度、相關性強度和關聯結構形式的不同，以致估計能力有別。 / The problem of estimating marginal distributions in a competing risks study is often met in scientific fields. Because main event and secondary event compete with each other, and a first occurring event prevents us from observing another event promptly, the intact lifetimes or survival times are unable to be collected in the circumstances that the probability of both lifetimes coinciding is 0. Unless lifetimes being independent or adding other conditions, there is a problem that the marginal distributions are non-identifiable. But the condition of independence is not always reasonable, we may assume the relation between lifetimes has some special form Because the copula defines the association between two variables, it can be employed to explain relation between lifetimes. Assuming that the dependence parameter in the copula framework is known, and adopting the concept of the probability integral transformations, this thesis has demonstrated whether the estimating abilities of the copula-graphic estimator, that Zheng and Klein put forward, are different in rates of censoring, intensities of dependence, and forms of the copula.

This article proposes the algorithm of a new control chart (loss function control chart) based on the Taguchi loss function with an adaptive scheme for binomial data. The loss function control chart is able to monitor cost variation from the process by applying loss function in the design. This new angle economically explores production cost. This research provides designs of the loss function control chart with specified VSI, optimal VSI, VSS and VP, respectively. Numerical analyses show that the specified VSI loss function chart, the optimal VSI loss function chart, the optimal VSS loss function chart and the optimal VP loss function chart outperform the Fp loss function chart significantly and show costs can be controlled systematically.

中文摘要 在本文中，我們依據模擬的風險因子變動，包括死亡率風險，利率風險，解約率風險以及模型的參數風險，來估計第一個保單年度的期末責任準備金之涉險值 (Value at Risk)。本文中，雖僅計算生死合險保單的準備金之涉險值，但是本文所提供的方法以及計算過程可以很容易的應用到其它險種，甚至配合資產面的考量來計算保險公司盈餘(Surplus)的涉險值，進而作為清償能力的監測系統。 本文的特點包括下列幾項：第一，本文提供了一個不同於傳統短期間(Short Horizon)的涉險值計算方式，來估計壽險商品的保單責任準備金(Policy Reserve)的涉險值。第二，本文利用生命表來估計死亡率風險所造成的涉險值。第三，我們利用隨機利率模型來捕捉隨機利率對於責任準備金涉險值的影響。第四，我們考慮解約率對於責任準備金涉險值的影響，值得注意的是，在我們的解約率模型中，引入的利率對於解約率的影響。第五，本文亦考慮風險因子模型當中的參數風險對於涉險值的影響。最後，我們利用無母數方法計算出涉險值的信賴區間，而信賴區間的估計在模擬過程當中尤其重要，因為它可以用來決定模擬次數的多寡。 本文包含六節：第一節為導論。第二節為計算死亡率風險的責任準備金涉險值。第三節是計算加上利率風險後責任準備金涉險值的變化。第四節則為加上解約率後對涉險值的影響。第五節為計算涉險值的信賴區間。第六節是我們的結論以及後續研究的方向探討。 本文包含六節：第一節為導論。第二節為計算死亡率風險的責任準備金涉險值。第三節是計算加上利率風險後責任準備金涉險值的變化。第四節則為加上解約率後對涉險值的影響。第五節為計算涉險值的信賴區間。第六節是我們的結論以及後續研究的方向探討。 / ABSTRACT In this paper, we estimate the VAR of life insurer's terminal reserve of the first policy year by the simulated risk factors, including mortality risk, interest rate risk, lapse rate risk, and estimation risks, of future twenty years. We found that the difference between the VAR under the mortality risk and the interest rate risk is very large because interest rate is a stochastic process but not mortality rate. Thus, the dispersion of interest rate is more then mortality rate. In addition, the VAR will reduce a lot after adding the impact of lapses because the duration of the reserve reduced. If we neglect the impact of lapses to VAR, we will overestimate the VAR significantly. The features of this paper are as follows. First, we provide an approach to measure the VAR of a life insurer's reserve, and it is rather different from traditional VAR with short horizons. Second, we use mortality table to estimate the VAR of a life insurer's reserve. Third, we use stochastic interest rate model to capture the effect of random interest rate to the VAR of a life insurer's reserve. Fourth, we relate the future cash outflows to interest rate and produce a reasonable estimator of VAR. Fifth, we consider the effect of estimation errors to the VAR of a life insurer's reserve. Last, we calculate the confidence interval of the VAR estimates of the policy reserves. This paper consists of six sections. The first section is an introduction. In the second section, we present the method used to estimate the variance of the mortality rate and then estimate the VAR of reserves from these variances. In the third section, we explore how to use stochastic interest rate model to estimate the reserve's VAR and the VAR associated with the parameter risk of the interest rate model. In the fourth section, we analyze the contribution of the lapse rate risk and the parameter risk of the lapse rate model to the reserve's VAR. We also analyze the relative significance of the interest rate risk, the lapse rate risk, and the mortality rate risk in terms of their marginal contributions to the VAR of an insurer's reserves in this section. In the fifth section, we calculate the confidence intervals of the VAR estimates discussed in the previous sections. The last section is the conclusion section containing our conclusions and discussions about potential future researches.