路燈巡修作業的落實與否，影響路燈維護的效率及品質，為能有效提升路燈管理之效能，近來管理階層逐漸重視路燈巡修區域的規劃。然而巡修區域的劃分，多依據主管人員之經驗調派，缺乏系統化、科學化的分析與評估，往往使人力資源無法有效運用，形成勞逸不均的現象，進而影響維護品質，因此如何以科學的方法劃分路燈巡修區域是個重要的課題。 本研究的重點在於針對現行路燈巡修區域劃分之缺點，提出一個新的方法，使各區域管理員巡修時間差達到最小化，以解決現行區域劃分的不合理現象。我們所提出的劃分法，以基因演算法進行演算，並加入禁忌名單改善基因演算法區域搜尋效率不佳的缺點，提升整體的求解速度，同時將路燈維護數量、故障率、維護時間、交通時間、巡修次數等影響因子，納入巡修時間的計算公式中，使劃分後各區的巡修時間差達到最小化。 本研究以台北市政府公園路燈工程管理處的路燈東區分隊為實作對象，在考慮不同的基因演化條件下，分別比較巡修區域劃分前後之變化情況，由實驗結果顯示，我們提出的劃分方法，確實使各區管理員的巡修時間差不超過3%，並且滿足巡修不跨區作業之需求。 / The efficiency and quality of street lights maintenance is influenced by the operation of patrolling and maintaining. In order to raise the working efficiency of maintenance crew, the supervisors pay more attention to region redistricting recently. The formor region districting methods normally base on human experiences without systematic or scientific evaluations, These facts, not only result in human resources wasting and uneven labor allocations, but also affecting the maintenance qualities. Therefore, it is a crucial issue to make region redistricting more scientifically. The key point of this research is to provide a systematic redistricting mechanism to minimize the patrolling time variation for all the districts. Our mechanism is based on genetic algorithm to reduce the patrolling time differences. Tabu search list is used to improve the searching efficiency of general genetic algorithms. Various factors were integrated in our mechanism to minimize the patrolling time variations. These factors include total number of street lights, average failure rate, average maintenance time, traffic delay, patrolling and maintaining frequency, etc. We used districts covered by the East Branch of SET/PSO of Taiepi City Government as the examples in our studies. The experimental results show that, using our mechanism, the patrolling time difference is reduced to 3% and maintenance crews can perform their duty without crossing region boundary.

近來共享經濟商業模式崛起，對各國既有相關市場皆造成不少之衝擊，當中，優步公司用以計算車資之「訂價演算法」，於美國實務亦引起許多爭議。美國司法案例中其中一個重要爭議即為優步公司單方制定之「訂價演算法」與其採行之「高峰動態訂價法」究否構成價格聯合行為。於美國實務近來2起與價格聯合行為相關之案例，即包含Meyer v. Kalanick案與Chamber of Commerce & RASIER, LLC v. City of Seattle案（以下簡稱「City of Seattle案」）中，皆可見Uber企圖正當化其價格聯合行為，以免於競爭法審查下有違法之嫌。而美國對於價格聯合行為之規範，載明於休曼法第1條；依據休曼法第1條規定，若原告擬主張被告行為違反卡特爾行為，則應證明系爭卡特爾行為符合合意主體要件、具合意或共謀行為，與造成限制性之競爭效果等三項要件。由於上述二案皆仍於訴訟前階段，判決尚未出爐，因此，此議題值得吾等分析之。本文擬以美國實務判決為基準，彙整相關爭議，進而探討Uber所採訂價演算法是否構成價格聯合行為。 本文發現，雖然此等訂價演算法究否構成價格聯合行為尚未有定論，然由於訂價演算法中之高峰動態訂價法可提高駕駛於尖峰時段中提供載客服務之誘因，將有助於調節市場機制與促進競爭。此外，Uber亦可利用其訂價演算法與設置平台所奠立之優勢，使其得以潛在破壞市場秩序之形式，創造競爭優勢。據此，Uber除可克服既有行政管制下市場進入之劣勢外，亦得使相關市場交易效率大幅提升、市場更加競爭。因此，於探討Uber價格聯合行為合法與否時，亦應將此等因素納入考量。 / The rapid expansion of sharing economy enterprises around the world has led to many challenges. And among these enterprises, one of the most disruptive examples is Uber because of its algorithm. In the United States, the lawsuits regarding Uber's algorithm has also gained massive attention. One of the controversial issues of the complaints relies upon whether Uber's algorithm which set by Uber, and “surge pricing” model do constitute an illegal price-fixing in violation of Section 1 of the Sherman Act. In 2 recent high-profile cases, Meyer v. Kalanick & Chamber of Commerce & RASIER, LLC v. City of Seattle, Uber has tried to justify its price fixing to avoid antitrust scrutiny. There are three specific facts that the Plaintiff must prove to establish its antitrust claim in Section 1 of the Sherman Act: 2 or more entities entering into an agreement, conspiracy, and unreasonably restrains competition. Analysis regarding Uber's algorithm is significant because the trials are ongoing. Therefore, the thesis examines whether Uber's algorithm do constitute an illegal price-fixing in violation of Section 1 of the Sherman Act by exploring the potential problems with regard to the elements based on U.S. judicial decisions. The thesis believes that Uber's algorithm can enhance the efficiency of transaction and has pro-competitive effects, leading to the impact of Uber's surge pricing on providing the incentives for drivers during peak hours. Establishing platform and Uber's algorithm create Uber's strengths and advantages. By having disrupted the existing industry, Uber's algorithm serves pro-competitive purposes.

隨著物聯網這個萬物連網的概念順勢推動智慧家庭在市場裡蓬勃發展，可預期未來ISP(Internet Service Provider)業者勢必面臨大量智慧家庭中各種不同應用服務互相競爭頻寬資源的情況，甚至遇到網路滿載壅塞時造成應用服務不堪使用的情形。 為改善上述問題，本文以ISP業者管理智慧家庭中眾多的物聯網設備為情境，透過軟體定義網路 (Software Defined Network，SDN)進行頻寬排程配置，排程演算法以可兼顧公平性(fairness)、時間延遲(delay)及應用服務優先權(service priority)的A-MLWDF (Adaptive Modified Largest Weighted Delay First) [7]演算法，確保優先配置頻寬給智慧家庭中優先權較高、時效較為急迫的流量，以降低應用服務的延遲來提升智慧家庭網路之服務品質(Quality of Service，QoS)。 本研究透過OMNet++模擬器建構SDN環境與傳統環境中有眾多物聯網設備之智慧家庭。家中物聯網設備包含M2M (Machine to Machine)和非M2M(non Machine to Machine)裝置，以提供各種智慧家庭應用服務。我們透過SDN架構進行頻寬配置，達到集中式管控家中的頻寬資源，其中排程演算法包括PF、MLWDF、A-MLWDF。實驗結果顯示，以上排程演算法雖然於SDN環境下在公平性與抖動率表現並不顯著，公平性約改善1.6%及抖動率約降低1%左右，但在產能與延遲方面表現較為顯著，能有效提高產能約52%，及降低延遲約 52%。 / With the concept of IoT (Internet of Things) spread rapidly, it is the opportunity to promote smart homes in the expanding market. We can see that the future ISP (Internet Service Provider) has to face a large number of smart homes having bandwidth competition in a variety of different applications and causing application services unavailable due to network congestion.     In order to resolve the above problems, we propose that each ISP (Internet Service Provider) has to manage a large number of IoT devices in a smart home to performs bandwidth scheduling through Software Defined Network (SDN). We choose to use A-MLWDF scheduling algorithm (Adaptive Modified Largest Weighted Delay First) [7] which considers fairness, delay and service priority. A-MLWDF is able to ensure services of higher priority and emergent traffic be allocated bandwidth earlier and greatly reduce delay and thus effectively enhance Quality of Service (QoS) of smart homes.     In this research, we implement a SDN environment by using OMNet++ to simulate the bandwidth competition among smart homes with IoT devices. The IoT devices consists of M2M (Machine to Machine) and non-M2M (non Machine to Machine) devices which offer a variety of intelligent home application services. We configure the bandwidth allocation under SDN control. The scheduling algorithms include PF, MLWDF and A-MLWDF. When the network traffic is congested, SDN can significantly increase throughput and reduce latency compared to traditional network management. The experimental results show that above scheduling algorithms using SDN environment having no significant performance improvements in fairness and jitter. The fairness increases around 1.6% and the jitter reduces around 1%. However, it shows significant improvement on throughout and delay. The throughput increases around 52% and the delay reduces around 52%.

本文以全球第一個演算法交易雲端平台－Quantopian進行研究，藉由平台社群討論區內公開之演算法交易策略，透過交易策略篩選和初步優化，以演算法交易策略為投資標的，搭配不同權重策略建構投資組合。權重策略部分，本文提出適用於組合式交易策略的績效指標加權 (Performance Index Weighted) 法，應用因子投資的觀念，融合排序相關性較低、不同面向之績效指標作為報酬率驅動因子，並參考Asness et al. (2013) 以因子排序作為權重計算依據，提供了簡單直覺、非最適化求解而且穩健的加權方式，更直接地將交易策略各面向績效的優劣反應在權重上。 根據數值分析，發現組合式交易策略長期而言，整體績效表現平均優於個別演算法交易策略，最小變異、績效指標加權和均等權重投資組合的風險亦明顯低於個別交易策略，且最小變異、績效指標加權和均等權重投資組合在降低投資組合風險的同時，並未犧牲過多報酬，風險調整後績效表現優於個別交易策略。而績效指標加權投資組合之年化報酬率、風險衡量和風險調整後績效表現皆優於最小變異、平均數－變異數、均等權重的加權投資組合，此種權重策略可使投資組合之夏普比率 (Sharpe ratio) 顯著提升，且投資組合的風險大幅降低，最大跌幅 (Max drawdown) 在四年半的實驗區間內降至10%以下的水準，風險調整後績效優異。 透過Quantopian社群演算法交易平台，個人投資者也能站在巨人的肩膀上學習，集合眾人的力量，憑藉量化交易創造出和機構法人一樣具有競爭力的投資組合。如Chan (2009) 所言，個人投資者也能憑藉量化交易，設計一套演算法交易策略。 / Quantopian is a crowd-sourced hedge fund which allows members on the platform to develop their own algorithmic strategies and even get capital allocations from Quantopian. In this paper, we constructed portfolios by Quantopian trading platform and proposed Performance Index Weighted method which generate consistently profit in our study. First, we filtered algorithmic trading strategies shared on the Quantopian community and improved the performance slightly. Second, we combined multiple algorithmic strategies with varied portfolio weight method, such as minimize-variance, performance index weighted, mean-variance, and equal weighed method to construct a portfolio. To elaborate, Performance Index Weighted portfolio is actually an application of factor investing, in which the portfolio weight depends on the ranking of performance index (factors), and these index measure returns, risk, and also risk-adjusted returns, which truly reflects how well the algorithmic strategy is. As a result, we used the performance index as a return driver and invested more in well-ranked strategies directly. Performance index weighted is a simple, robust, and fully intuitively way to construct a portfolio. In numerical analysis, we found that using multiple strategies to construct a portfolio could generate better performance than a single algorithm strategy on average. Moreover, the annual returns, risk measure, and risk-adjusted returns of Performance Index Weighted portfolio turn out to be better than minimize-variance portfolio, mean-variance portfolio, and equal weighted portfolio. As a result, Performance Index Weighted portfolio has significantly higher Sharpe ratio and lower Max Drawdown (lower than 10% in our out-of-sample test) than other portfolios, which shows excellent risk-adjusted performance. Most important of all, retail traders could learn more precisely by standing on the shoulders of giants through Quantopian trading platform. Also, by collecting wisdom from the crowd, we create an opportunity for retail traders to construct competitive portfolios just as institutional investors do.

本研究利用美國芝加哥商品交易所針對 18 個城市發行之冷氣指數／暖氣指數衍生性商品與相對應之日均溫進行分析與評價。研究成果與貢獻如下：一、延伸 Alaton, Djehince, and Stillberg (2002) 模型，引入跳躍風險、隨機波動度、波動跳躍等因子，提出新模型以捕捉更多溫度指數之特徵。二、針對不同模型，分別利用最大概似法、期望最大演算法、粒子濾波演算法等進行參數估計。實證結果顯示新模型具有較好之配適能力。三、利用 Esscher 轉換將真實機率測度轉換至風險中立機率測度，並進一步利用　Feynman-Kac 方程式與傅立葉轉換求出溫度模型之機率分配。四、推導冷氣指數／暖氣指數期貨之半封閉評價公式，而冷氣指數／暖氣指數期貨之選擇權不存在封閉評價公式，則利用蒙地卡羅模擬進行評價。五、無論樣本內與樣本外之定價誤差，考慮隨機波動度型態之模型對於溫度衍生性商品皆具有較好之評價績效。六、實證指出溫度市場之市場風險價格為負，顯示投資人承受較高之溫度風險時會要求較高之風險溢酬。本研究可給予受溫度風險影響之產業，針對衍生性商品之評價與模型參數估計上提供較為精準、客觀與較有效率之工具。 / This study uses the daily average temperature index (DAT) and market price of the CDD/HDD derivatives for 18 cities from the CME group. There are some contributions in this study: (i) we extend the Alaton, Djehince, and Stillberg (2002)'s framework by introducing the jump risk, the stochastic volatility, and the jump in volatility. (ii) The model parameters are estimated by the MLE, the EM algorithm, and the PF algorithm. And, the complex model exists the better goodness-of-fit for the path of the temperature index. (iii) We employ the Esscher transform to change the probability measure and derive the probability density function of each model by the Feynman-Kac formula and the Fourier transform. (iv) The semi-closed form of the CDD/HDD futures pricing formula is derived, and we use the Monte-Carlo simulation to value the CDD/HDD futures options due to no closed-form solution. (v) Whatever in-sample and out-of-sample pricing performance, the type of the stochastic volatility performs the better fitting for the temperature derivatives. (vi) The market price of risk differs to zero significantly (most are negative), so the investors require the positive weather risk premium for the derivatives. The results in this study can provide the guide of fitting model and pricing derivatives to the weather-linked institutions in the future.

在音樂中，組曲是一種特別的創作形式。組曲將多首音樂段落組合排列，並且在音樂段落之間加入間奏，形成一首音樂組曲。組曲的編曲重點在於音樂段落的編排順序及段落之間的連結。平時在宴會、舞會、餐廳、賣場等場合中，往往都會連續播放多首流行音樂。利用電腦編曲自動產生流行音樂組曲，將可提升播放音樂的銜接與流暢感。 因此，本研究利用資料探勘技術及音樂編曲理論，將多首音樂重新改編成一首組曲。系統首先將每首音樂分段並找出每首音樂的代表段落。接著，系統根據代表段落間的相似度編排順序。最後，為了達到組曲中音樂段落連接的流暢性，我們以模型訓練的方式在段落連結間加入間奏。系統從訓練資料學習產生旋律發展、和弦進程與節奏的模型，接著分析代表段落的動機、旋律、和弦及節奏，使得組曲編曲後的段落連結更為流暢且完整。本研究以流行音樂鋼琴伴奏曲為測試資料，我們分別邀請三十四位受過音樂訓練與未受音樂訓練的測試者，針對本研究所提出的鋼琴伴奏節奏辨識、代表段落萃取、段落順序編排及間奏產生，評估其效果。實驗結果顯示，本研究所提出的順序編排與間奏產生技術，對於組曲的流暢感，有著相當大的幫助。 / In music, a medley is an organized piece composed from segments of existing pieces. Ordering and bridge for connection between segments are the important elements for medley arrangement. Automatic medley arrangement is helpful for playing a set of music continuously which is common in banquet, party, restaurant, shopping mall, etc.. This thesis aims to develop the automatic medley arrangement method by using data mining techniques and music arrangement theory. The proposed method first segments each music and discovers the significant segment from each music. Then, the linear arrangement based on the similarities between significant segments is generated. Finally, in order to connect segments smoothly in the medley, the bridge between two segments is generated and inserted by using model training. Three models, melody progression, chord progression and rhythm models are learned from training data. For the experiments, testing data is collected from popular piano music and thirty-four people are invited to evaluate the effectiveness of the rhythm recognition of accompaniment, the extraction of significant segment, the linear arrangement of segments, and the creation of bridge. Experimental results show that the proposed medley arrangement method performs well.

在無線網狀網路(Wireless Mesh Network)中，每個節點須幫助相鄰節點轉送資料及提供使用者網路存取，例如WLAN(IEEE 802.11s)、WMAN(IEEE 802.16)等，皆可利用多跳接方式將資料轉送至通訊閘道器(Gateway)。在無線網狀網路中，常利用密集佈建的方式來解決通訊死角的問題。當網路節點的密度增加時，無線訊號的干擾也會增強，並且各節點的效能會顯著下降。 在本研究中，將利用幾何學概念，解決網路干擾問題，並提出拓樸重建演算法來重建路徑，使網路干擾達到最小化。我們試著最小化節點與節點間的干擾，以提升整體無線網狀網路效能。我們將網路問題轉換成幾何問題，並定義在幾何圖形中線段交錯問題，之後驗證在幾何圖形中是否有線段交錯的現象發生。若發生線段交錯時，則將此線段從幾何圖形中移除，並且利用三角化演算法將此區域線段重新規劃，使相鄰節點間的干擾最小。當網路拓樸建立完成後，我們利用標準差公式將干擾較大的連線移除，使得網路效能提升。上述測試線段交錯及三角化多邊形演算法可在時間複雜度O(n log n)內找到干擾最小的解。最後，我們將利用網路模擬器(Network Simulator)驗證所提出的方法是否能達到預期的系統效能指標。 / In wireless mesh networks, such as WLAN (IEEE 802.11s), WMAN (IEEE 802.16), etc., each node should forward packets of neighboring nodes toward gateway using multi-hop routing mechanism. In wireless mesh network, as the density of network nodes increases, the RF interference will increase and the throughput of each node will drop rapidly. In our research, we use the geometry to resolve the RF interference problem by rebuilding network topology. We try to minimize the interference between neighboring nodes and improve the throughput in wireless mesh network. We transform the network topology problem into geometry problem and define the line intersection problem in geometric graph, then check path intersection in the geometric graph. If line intersection occurs in the graph, we remove the intersection line from the graph and re-plan the region by triangulation algorithm. When the network topology is built up, we use a standard deviation formula to improve network performance by removing longer links. The line intersection algorithm and triangulation algorithm, both of time complexity O(n log n), are used to find the minimal interference solution. At the end of our research, we use network simulator to verify if the proposed methods can help to meet all those performance expectations.

本研究將探討非等效平行機台問題中具備順序相依整備時間及不同開始工作時間（Unequal ready-time）之情況，並以最小化總延遲工件權重數為目標值，其目的在改善非等效平行機台問題應用於實際產業中製造環境裡所面對的各項挑戰，如印刷電路板的鑽孔和半導體的測試製程。因本研究欲求解之問題是屬於NP - Hard problems 性質之尋優問題，故利用啟發式方法(heuristics)求解為合適的選擇。此外，本研究計畫開發混合式巨集啟發式方法來求解非等效平行機台問題，主要以禁忌搜尋法為主，在鄰域的搜尋上，也藉由變動鄰域尋優法能夠透過鄰域轉換的機制，進而找出更多好的解。由於啟發式方法對於尋優問題常需花費許多時間來計算才能獲得更好的解，為確保增進求解效率與品質，將針對問題特性開發數種初始解產生法，並也嘗試定義幾個能夠減少尋找鄰近解之鄰域。在後續求解改善的過程中，主要整合變動鄰域(VND)及禁忌(TS)巨集啟發式演算法搜尋最佳解。此外，為了評估本文推導之演算法效能，本研究利用設定之條件隨機產生適量模擬現場狀況的測試情境，進而比較本研究所提出之混合式巨集啟發式方法及標準禁忌搜尋法在不同情境下之表現。 / The problem considered in this paper is a set of independent jobs on unrelated parallel machines with sequence-dependent setup times and with unequal ready times so as to minimize total weighted tardy jobs. These problems can be found in real-life manufacturing environments, such as PCB fabrication drilling operations and semiconductor wafer manufacturing dicing. Since the problems are NP-hard in the strong sense, heuristics are an acceptable practice to finding good solutions. A hybrid meta-heuristics are proposed to solve this scheduling problem. The proposed heuristics belong to a type of solution improvement heuristic; therefore, the heuristics start with an effective initial feasible solution then a meta-heuristic is applied to improve the solution. To enhance both the efficiency and efficacy of the heuristics, several different initial solution generators, based on the characteristics of problems, are developed. The meta-heuristic is a hybrid heuristic integrating the principles of Variable Neighborhood Descent approach (VND) and Tabu Search (TS). In order to evaluate the performance of the proposed heuristics, two sets of large number test scenarios will be designed to simulate practical shop floor problems. Computational experiments will be performed to compare the performance of the proposed heuristics, and a basic tabu search algorithm. The results show the proposed heuristic perform better than the basic tabu search algorithm.

在傳統的線性迴歸分析當中，當基本假設不滿足時，有時可考慮變數轉換使得資料能夠比較符合基本假設。在眾多的轉換方法當中，以Box和Cox(1964)所提出的乘冪轉換(Box-Cox power transformation)最為常用，乘冪轉換可將某些複雜的系統轉換成線性常態模式。然而當資料存在離群值(outlier)時，Box-Cox Transformation會受到影響，因此不是一種穩健方法。 在本篇論文當中，我們利用前進演算法(forward search algorithm)求得最小消去平方估計量(Least trimmed squares estimator)，在過程當中估計出穩健的轉換參數。

在處理具遺失值之類別資料時，傳統的方法是將資料捨棄，但是這通常不是明智之舉，這些遺失某些分類訊息的資料通常還是可以提供其它重要的訊息，尤其當這類型資料的個數佔大多數時，將其捨棄可能使得估計的變異數增加，甚至影響最後的決策。如何將這些遺失某些訊息的資料納入考慮，作出完整的分析是最近幾十年間頗為重要的課題。本文主要整理了五種分析這類型資料的方法，分別為單樣本方法、多樣本方法、概似方程式因式分解法、EM演算法，以上四種方法可使用在資料遺失呈隨機分佈的條件成立下來進行分析。第五種則為樣本遺失不呈隨機分佈之分析方法。 / Traditionally, the simple way to deal with observations for which some of the variables are missing so that they cannot cross-classified into a contingency table simply excludes them from any analysis. However, it is generally agreed that such a practice would usually affect both the accuracy and the precision of the results. The purpose of the study is to bring together some of the sound alternatives available in the literature, and provide a comprehensive review. Four methods for handling data missing at random are discussed, they are single-sample method, multiple-sample method, factorization of the likelihood method, and EM algorithm. In addition, one way of handling data missing not at random is also reviewed.