In view of the current situation of fluctuating high crude oil prices, it is now more important than ever for petroleum refineries to operate at an optimal level in the present dynamic global economy. Acknowledging the shortcomings of deterministic models, this work proposes a hybrid of stochastic programming formulations for an optimal midterm refinery planning that addresses three factors of uncertainties, namely price of crude oil and saleable products, product demand, and production yields. An explicit stochastic programming technique is utilized by employing compensating slack variables to account for violations of constraints in order to increase model tractability. Four approaches are considered to ensure both solution and model robustness: (1) the Markowitz’s mean–variance (MV) model to handle randomness in the objective coefficients of prices by minimizing variance of the expected value of the random coefficients; (2) the two-stage stochastic programming with fixed recourse approach via scenario analysis to model randomness in the right-hand side and left-hand side coefficients by minimizing the expected recourse penalty costs due to constraints’ violations; (3) incorporation of the MV model within the framework developed in Approach 2 to minimize both the expectation and variance of the recourse costs; and (4) reformulation of the model in Approach 3 by adopting mean-absolute deviation (MAD) as the risk metric imposed by the recourse costs for a novel application to the petroleum refining industry. A representative numerical example is illustrated with the resulting outcome of higher net profits and increased robustness in solutions proposed by the stochastic models.

Sample average approximation (SAA) is a well-known solution methodology for traditional stochastic programs which are risk neutral in the sense that they consider optimization of expectation functionals. In this thesis we establish sample average approximation methods for two classes of non-traditional stochastic programs. The first class is that of stochastic min-max programs, i.e., min-max problems with expected value objectives, and the second class is that of expected value constrained stochastic programs. We specialize these SAA methods for risk-averse stochastic problems with a bi-criteria objective involving mean and mean absolute deviation, and those with constraints on conditional value-at-risk. For the proposed SAA methods, we prove that the results of the SAA problem converge exponentially fast to their counterparts for the true problem as the sample size increases. We also propose implementation schemes which return not only candidate solutions but also statistical upper and lower bound estimates on the optimal value of the true problem. We apply the proposed methods to solve portfolio selection and supply chain network design problems. Our computational results reflect good performance of the proposed SAA schemes. We also investigate the effect of various types of risk-averse stochastic programming models in controlling risk in these problems.

Penalized methods are becoming more and more popular in statistical research. This dissertation research covers two major aspects of applications of penalized methods: variable selection and nonparametric function estimation. The following two paragraphs give brief introductions to each of the two topics. Infinite variance autoregressive models are important for modeling heavy-tailed time series. We use a penalty method to conduct model selection for autoregressive models with innovations in the domain of attraction of a stable law indexed by alpha is an element of (0, 2). We show that by combining the least absolute deviation loss function and the adaptive lasso penalty, we can consistently identify the true model. At the same time, the resulting coefficient estimator converges at a rate of n^(?1/alpha) . The proposed approach gives a unified variable selection procedure for both the finite and infinite variance autoregressive models. While automatic smoothing parameter selection for nonparametric function estimation has been extensively researched for independent data, it is much less so for clustered and longitudinal data. Although leave-subject-out cross-validation (CV) has been widely used, its theoretical property is unknown and its minimization is computationally expensive, especially when there are multiple smoothing parameters. By focusing on penalized modeling methods, we show that leave-subject-out CV is optimal in that its minimization is asymptotically equivalent to the minimization of the true loss function. We develop an efficient Newton-type algorithm to compute the smoothing parameters that minimize the CV criterion. Furthermore, we derive one simplification of the leave-subject-out CV, which leads to a more efficient algorithm for selecting the smoothing parameters. We show that the simplified version of CV criteria is asymptotically equivalent to the unsimplified one and thus enjoys the same optimality property. This CV criterion also provides a completely data driven approach to select working covariance structure using generalized estimating equations in longitudinal data analysis. Our results are applicable to additive, linear varying-coefficient, nonlinear models with data from exponential families.

In this study within quantitative portfolio optimization, stochastic programming is investigated as an investment decision tool. This research takes the direction of scenario based Mean-Absolute Deviation and is compared with the traditional Mean-Variance model and widely used Risk Parity portfolio. Furthermore, this thesis is done in collaboration with the First Swedish National Pension Fund, AP1, and the implemented multi-asset portfolios are thus tailored to match their investment style. The models are evaluated on two different fund management levels, in order to study if the portfolio performance benefits from a more restricted feasible domain. This research concludes that stochastic programming over the investigated time period is inferior to Risk Parity, but outperforms the Mean-Variance Model. The biggest aw of the model is its poor performance during periods of market stress. However, the model showed superior results during normal market conditions. / I denna studie inom kvantitativ portföljoptimering undersöks stokastisk programmering som ett investeringsbeslutsverktyg. Denna studie tar riktningen för scenariobaserad Mean-Absolute Deviation och jämförs med den traditionella Mean-Variance-modellen samt den utbrett använda Risk Parity-portföljen. Avhandlingen görs i samarbete med Första AP-fonden, och de implementerade portföljerna, med era tillgångsslag, är därför skräddarsydda för att matcha deras investeringsstil. Modellerna utvärderas på två olika fondhanteringsnivåer för att studera om portföljens prestanda drar nytta av en mer restrektiv optimeringsmodell. Den här undersökningen visar att stokastisk programmering under undersökta tidsperioder presterar något sämre än Risk Parity, men överträffar Mean-Variance. Modellens största brist är dess prestanda under perioder av marknadsstress. Modellen visade dock något bättre resultat under normala marknadsförhållanden.

指數型基金透過追蹤指數來提供投資人被動管理的投資策略，因而成為保守投資人的熱門投資工具。本論文的目的在於建立一個追蹤指數的同時也能有效控管損失的指數型基金。在此目標下，該基金面臨到的不單是追蹤指數的績效，還有降低資產配置風險的問題。有鑑於此，本論文融合兩種下方風險的概念：指數追蹤的下方偏差(downside absolute deviation)以及條件風險值(conditional value-at-risk, CVaR)。針對兩者間的規避程度分別分配其權重，並以該基金之平均報酬大於指數的平均報酬作為限制條件，經由改寫下方偏差與離散化CVaR後得到一個新的線性規劃模型。本論文以台灣50指數與恆生指數的歷史資料做為實證探討的對象，驗證使用本線性規劃模型所建立之指數型基金的效能。 / Index fund has become popular in these days among the conservative investors since it provides a passive investment strategy. The main purpose of this paper is to create an index fund which can replicate the performance of a broad-based index of stocks and has the ability to control the loss efficiently at the same time. For this purpose, the index fund we build confronts with not only the performance of index tracking, but also lowering the level of the risk of assets allocation. In order to accomplish the goal, we combine two concepts of downside risk: downside absolute deviation and conditional value-at-risk (CVaR). Under the constraint of average portfolio return being greater than average index return, and assign weights according to the degree of evasion to each of the risks, a linear programming model is formulated by rewriting downside absolute deviation and discretizing CVaR. The results obtained from the computational experience on Taiwan 50 index and Hang Seng index are provided for testing the efficiency of this model.

The ordinary least squares (OLS) is one of the most widely used methods for modelling the functional relationship between variables. However, this estimation procedure counts on some assumptions and the violation of these assumptions may lead to nonrobust estimates. In this study, the simple linear regression model is investigated for conditions in which the distribution of the error terms is Generalised Logistic. Some robust and nonparametric methods such as modified maximum likelihood (MML), least absolute deviations (LAD), Winsorized least squares, least trimmed squares (LTS), Theil and weighted Theil are compared via computer simulation. In order to evaluate the estimator performance, mean, variance, bias, mean square error (MSE) and relative mean square error (RMSE) are computed.

隨著WiFi訊號在都市的涵蓋率逐漸普及，基於WiFi訊號強度值的定位方法逐漸發展。WiFi匹配定位(Matching Positioning)是透過參考點坐標與WiFi訊號強度(Received Signal Strength Indicator, RSSI)的蒐集，以最小二乘法(Least Squares, LS)計算RSSI模型參數；然後，利用模型參數與使用者位置的WiFi訊號強度，推估出使用者的位置。然而WiFi訊號強度容易受到環境因素影響，例如降雨、建物遮蔽、人群擾動等因素，皆會使訊號強度降低，若以受影響的訊號強度進行定位，將使定位成果與真實位置產生偏移。 為了降低訊號強度的錯誤造成定位結果的誤差，本研究嘗試透過具有穩健性的最小一乘法( Least Absolute Deviation, LAD)結合WiFi匹配定位，去克服WiFi訊號易受環境影響的特性，期以獲得較精確的WiFi定位成果。研究首先透過模擬資料的建立，測試不同粗差狀況最小一乘法WiFi匹配定位之表現，最後再以真實WiFi訊號進行匹配定位的演算，並比較最小一乘法WiFi匹配定位與最小二乘法WiFi匹配定位的成果差異，探討二種方法的特性。 根據本研究成果顯示，於模擬資料中，最小一乘法WiFi匹配定位相較於最小二乘法WiFi匹配定位，在面對參考點接收的AP訊號與檢核點接收的AP訊號強度含有粗差的情形皆能有較好的穩健性，且在參考點接收的AP訊號含有粗差的情況有良好的偵錯能力。而於真實環境之下，最小一乘法WiFi匹配定位之精度也較最小二乘法WiFi匹配定位具有穩健性；在室外資料的部份，最小一乘法WiFi匹配定位之精度為8.46公尺，最小二乘法WiFi匹配定位之精度為8.57公尺。在室內資料的部份，最小一乘法WiFi匹配定位之精度為2.20公尺，最小二乘法WiFi匹配定位之精度為2.41公尺。 / Because of the extensive coverage of WiFi signal, the positioning methods by the WiFi signal are proposed. WiFi Matching Positioning is a method of WiFi positioning. By collecting the WiFi signal strength and coordiates of reference points to calculate the signal strength transformation parameters, then, user’s location can be calculated with the LS (Least Squares). However, the WiFi signal strength is easily degraded by the environment. Using the degraded WiFi signal to positioning will produce wrong coordinates. Hence this research tries to use the robustness of LAD (Least Absolute Deviation) combining with WiFi Matching Positioning to overcome the sensibility of WiFi signal strength, expecting to make the result of WiFi positioning more reliable. At first, in order to test the ability of LAD, this research uses simulating data to add different kind of outliers in the database, and checks the performance of LAD WiFi Matching Positioning. Finally, this research uses real data to compare the difference between the results of LAD and LS WiFi Matching Positioning. In the simulating data, the test result shows that LAD WiFi Matching Positioning can not only have better robust ability to deal with the reference and check points AP signal strength error than LS WiFi Matching Positioning but also can detect the outlier in the reference points AP signal strength. In the real data, LAD WiFi Matching Positioning can also have better result. In the outdoor situation, the RMSE (Root Mean Square Error) of LAD WiFi Matching Positioning and LS (Least Squares) WiFi Matching Positioning are 8.46 meters and 8.57 meters respectively. In the indoor situation, the RMSE (Root Mean Square Error) of LAD WiFi Matching Positioning and LS (Least Squares) WiFi Matching Positioning are 2.20 meters and 2.41 meters respectively.

Wilcox (1990) examined the Type I and Type II error rates for several robust tests of H0: σ12/σ22 = 1 in paired-data designs and concluded that a satisfactory solution does not yet exist. A confidence interval for a ratio of correlated mean absolute deviations is derived and performs well in small sample sizes across realistically nonnormal distributions. When used to test a hypothesis, the proposed confidence interval is almost as powerful as the most powerful test examined by Wilcox.

本文就壽險業投資情況、投資組合與不動產市場的關聯性進行討論，介紹壽險業目前的營運狀況、資金來源以及資金運用的情況與限制等；並且試圖描述不動產市場之情形，說明壽險業可能進入不動產市場的契機、以及提出資金進入不動產市場的可能策略；其後建立新的不動產市場投資之變數，本文以不動產貸款抵押債券投入所建立的MIN－MAD﹙Mean-absolute Deviation﹚Model，瞭解投入此新變數後對投資組合的變化並比較其不同之處，藉此探究以其他方式將資金投入是否為一可行之策略。

Made available in DSpace on 2016-06-02T19:53:32Z (GMT). No. of bitstreams: 1 MARTINEZ_Jonathan_2014.pdf: 11989383 bytes, checksum: 96fb53d9544014ea55b1e53ee779c134 (MD5) Previous issue date: 2014-03-14 / Financiadora de Estudos e Projetos / This study presents a pump scheduling problem for the capture, transfer and storage of water supply systems in urban networks, whose objective is to minimize the electricity cost associated to the pumping operations. To deal with the dynamic and random nature of the water-demand, we propose two-stage stochastic programming with recourse models, where the random variables are represented by a finite and discrete set of realizations or scenarios. The developed mathematical models are extensions of previous deterministic models of the literature and they reflect the basic assumption that a fixed cost could be incurred by the turn on/ turn off activities of the hydraulic pumps. In order to control violations of the water-demand constraints in the presence of multiple different scenarios, we also consider a robustness technique in an attempt to obtain almost feasible solutions. Last, but not least, we adopt a risk-aversion criteria so-called mean absolute deviation to obtain second-stage costs less dependent on the realizations of the scenarios. The scenarios were generated according to a Monte-Carlo simulation procedure that may use any probability distributions to produce the empirical probabilities of the random variables. As the proposed pump scheduling problem with fixed cost is a two-stage stochastic mixed 0 − 1 program, we develop a efficient hybrid heuristic to obtain good-quality solutions of practical instances in a plausible running time. Overall results evidence the stability of the scenario generation method, the sensitivity of the solution according to the key parameters of the mathematical model, and the efficiency of the heuristic in solving large instances. Finally, we show that is possible to save resources by solving the stochastic programming model instead of adopting simpler approaches based on the expected value. / Esse estudo apresenta um problema de programação de bombas para a captação, armazenamento e transferência de água em sistemas de abastecimentos de água em redes urbanas, cujo objetivo é minimizar o custo de energia elétrica associado às operações de bombeamento. Para lidar com a natureza dinâmica e aleatória da demanda por água, foram propostos modelos de programação estocástica de dois estágios com recurso, em que a variável aleatória é representada por um conjunto finito de realizações ou cenários. Os modelos matemáticos desenvolvidos são extensões de modelos determinísticos da literatura e refletem a suposição básica de que é possível se incorrer em um custo fixo pelas atividades de liga/desliga das bombas hidráulicas. Para controlar as violações das restrições de demanda por água na presença de múltiplos cenários diferentes, considerou-se também uma técnica de robustez na tentativa de gerar soluções quase factíveis. Por último, mas não menos importante, adotou-se um critério de aversão ao risco denominado desvio médio absoluto para obter custos de segundo estágio menos dependentes das realizações dos cenários. Os cenários foram gerados de acordo com um procedimento baseado em simulação Monte-Carlo que pode utilizar qualquer distribuição de probabilidade para produzir as probabilidades empíricas das variáveis aleatórias. Como o problema de programação de bombas com custo fixo proposto é um programa inteiro misto 0−1 estocástico, desenvolve-se uma heurística híbrida eficiente para obter soluções de boa qualidade de instâncias práticas em um tempo computacional plausível. Os resultados evidenciam a estabilidade do método de geração de cenários, a sensibilidade da solução de acordo com parâmetros-chave do modelo matemático, e a eficiência da heurística na resolução de instâncias de grande porte. Finalmente, foi demonstrado que é possível poupar recursos pela resolução do modelo de programação estocástica, em vez de adotar abordagens mais simples baseadas no valor esperado.