Capable and well-organized data mining algorithms are essential and fundamental to helpful, useful, and successful knowledge discovery in databases. We discuss several data mining algorithms including genetic algorithms (GAs). In addition, we propose a modified multivariate Newton's method (NM) approach to data mining of technical data. Several strategies are employed to stabilize Newton's method to pathological function behavior. NM is compared to GAs and to the simplex evolutionary operation algorithm (EVOP). We find that GAs, NM, and EVOP all perform efficiently for well-behaved global optimization functions with NM providing an exponential improvement in convergence rate. For local optimization problems, we find that GAs and EVOP do not provide the desired convergence rate, accuracy, or precision compared to NM for technical data. We find that GAs are favored for their simplicity while NM would be favored for its performance.

Social media has provided a way for music artists to reach many people with their music, without having to rely on record labels to perform marketing tasks. Most previous research within the area has focused on how already established music artists can use social media as part of their marketing strategies and how digital technologies have transformed the music industry. This study focuses on how unsigned music artists’ followers and fans on social media have an impact on their music streaming numbers. The main research question of the study is: how does unsigned artists’ social media performance affect the number of streams of their music? To investigate this, a robust regression model was defined with the aim of predicting the number of artists’ music streams based on their social media data. The robust regression model showed that the social media variables did not have significant effects on the number of streams. Therefore, an analysis of each individual artist in the data was conducted. The results showed that the social media data in this study could not be used to explain changes in the number of streams for unsigned music artists. An analysis based on each individual artist and the content that each individual artist is posting on the different social media channels, is suggested instead. An information visualization tool was developed with the purpose of allowing analysts to get an overview of the social media data as well as allow analysts to look at each artist’s social media feeds to understand how artists’ social media activities affect their music streaming data. / Sociala medier har gjort det möjligt för musikartister att nå många människor med sin musik utan att behöva förlita sig på skivbolag. Tidigare forskning inom området har fokuserat på hur redan etablerade musikartister kan använda sociala medier som en del av sina marknadsstrategier och hur digital teknik har förändrat musikbranschen. Denna studie fokuserar på hur osignerade musikartisters antal anhängare och fans på sociala medier påverkar antalet streams av artisternas musik. Studiens huvudsakliga forskningsfråga är: Hur påverkar osignerade artisters prestationer på sociala medier antalet streams av deras musik? För att undersöka detta definierades en robust regressionsmodell i syfte att förutse antalet streams av artisternas musik baserat på deras sociala mediedata. Den robusta regressionsmodellen visade att socialamedievariablerna inte hade signifikanta effekter på antalet streams av artisternas musik. Därför genomfördes en analys av varje enskild artist i datan. Resultaten visade att sociala mediedatan i denna studie inte kunde användas för att förklara förändringar i antalet streams för osignerade musikartister. En analys baserad på varje enskild artist och innehållet som varje enskild artist lägger ut på de olika sociala mediekanalerna föreslås istället. Ett informationsvisualiseringsverktyg utvecklades med syftet att ge analytiker en möjlighet att få en överblick över sociala mediedatan samt låta analytiker titta på varje artists sociala medieflöden för att förstå hur artisternas sociala medier påverkar deras musikstreamingdata.

本文探討退休金發展程度與投資策略對股票市場發展的影響，並同時採用「縱橫門檻迴歸模型」（panel threshold model, PTM）及結合縱橫門檻模型與穩健迴歸的「穩健縱橫門檻迴歸模型」（robust panel threshold model, ROPTM）來研究此一議題。我們用退休基金投資證券市場的金額佔總額的比例為分類標準，將樣本分為高投資比例與低投資比例兩部分。對部分OECD國家及台灣的panel data分析後之結果顯示：在股票市場方面，若基金採高投資比例之投資策略，則退休金發展或投資股市比例越高，越能促進股市發展；採低投資比例策略的基金，對股市發展的影響並不顯著。 / This paper analyzes the impact of pension fund investment on securities markets using a panel threshold model (PTM) and a robust panel threshold model (ROPTM) which combines a panel threshold model with a robust regression model. We use panel data for some OECD countries and Taiwan to test the validity of our propositions. The data is divided into low and high investment regions based on the value of securities as a percentage of total financial assets of the pension fund. Our results are the following. In the high stock investment region, pension funds have a positive impact on stock markets. Whereas, in the low stock investment region, the positive impact seems to disappear.

This thesis deals with modern statistical approaches and their application aimed at robust methods and neural network modelling. Selected methods are analyzed and applied on frequent practical problems in czech industry and technology. Topics and methods are to be benificial in real applications compared to currently used classical methods. Applicability and effectivity of the algorithms is verified and demonstrated on real studies and problems in czech industrial and research bodies. The great and unexploited potential of modern theoretical and computational capacity and the potential of new approaces to statistical modelling and methods. A significant result of this thesis is also an environment for software application development for data analysis with own programming language DARWin (Data Analysis Robot for Windows) for implemenation of effective numerical algorithms for extaction information from data. The thesis should be an incentive for boarder use of robust and computationally intensive methods as neural networks for modelling processes, quality control and generally better understanding of nature.

在一般的研究過程中均可能有離群觀測值產生，只要有離群觀測值存在， 就可能對研究結果產生極重大的影響。在統計學上常用的參數估計式中， 有許多極易受離群觀測值影響。因此本研究採用模糊族群分析混合最大概 似估計演算法運用在參數估計上，以去除離群觀測值對分析結果的影響。 本研究主要針對相關係數與迴歸係數的估計進行探討，利用演算法中所求 得之隸屬度，計算穩健相關係數和穩健迴歸係數，以期能正確估計參數值 。

Submitted by ANA KARLA PEREIRA RODRIGUES (anakarla_@hotmail.com) on 2017-08-09T15:33:48Z No. of bitstreams: 1 arquivototal.pdf: 4699110 bytes, checksum: ef3b7c0aa5c4758d2c77e65ad6a81ad3 (MD5) / Made available in DSpace on 2017-08-09T15:33:48Z (GMT). No. of bitstreams: 1 arquivototal.pdf: 4699110 bytes, checksum: ef3b7c0aa5c4758d2c77e65ad6a81ad3 (MD5) Previous issue date: 2016-06-20 / Conselho Nacional de Pesquisa e Desenvolvimento Científico e Tecnológico - CNPq / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES / This work involves the development of two studies that are presented in chapters 2 and 3. At first, a new method to perform the calibration transfer was designed. This method was developed to make use of separate variables instead of using the full spectrum or spectral windows. To accomplish this task a univariate procedure is initially used to correct the spectra recorded in the secondary equipment, given a set of transfer samples. A robust regression technique is then used to obtain a model with small sensitivity with respect to the univariate correction. The proposed method is employed in two case studies involving near infrared spectrometric determination of specific mass, research octane number and naphtenes in gasoline, and moisture and oil in corn. In both cases, better calibration transfer results were obtained in comparison with piecewise direct standardization (PDS). In the second, a new strategy for cotton seed classification using near infrared (NIR) hyperspectral images (HSI) was developed. Initially the cotton seeds samples were recorded on a station HSI image-NIR and a conventional spectrometer NIR. Thereon, the images were segmented and the mean spectrum of each seed was extract. Classification models SPA-LDA e PLS-DA based on the mean spectral were developed for two data sets. The results for models SPA-LDA and PLSDA showed that the classification with HSI-NIR data set has been achieved with greater accuracy when compared to models for the NIR-conventional data set. / Este trabalho envolve o desenvolvimento de dois estudos, que são apresentados nos capítulos 2 e 3. No primeiro, um novo método para realizar a transferência de calibração foi concebido. Este método foi desenvolvido para fazer uso de variáveis isoladas em vez de usar todo o espectro ou janelas espectrais. Para realizar essa tarefa, um procedimento univariado é inicialmente usado para corrigir os espectros registrados no equipamento secundário, dado um conjunto de amostras de transferência. Uma técnica de regressão robusta é então usada para obter um modelo com pequena sensibilidade em relação aos resíduos da correção univariada. O novo método é então empregado em dois estudos de caso envolvendo análise espectrométrica NIR, em que foram determinados os parâmetros massa específica, RON (Research Octane Number) e teor de naftênicos em gasolina e os teores de água e óleo em amostras de milho. Os resultados do novo método foram melhores do que os obtidos usando o método PDS. No segundo, uma nova estratégia para classificação de sementes de algodão usando imagens hiperespectrais no NIR foi desenvolvido. Inicialmente as amostras de sementes de algodão foram registradas em uma estação de imagem HSI-NIR e em um equipamento NIR convencional. Após isso, as imagens foram segmentadas e os espectros médios de cada semente foram extraídos. Os modelos de classificação SPA-LDA e PLS-DA baseados nos espectros médios foram construídos para os dois conjuntos de dados. Os resultados SPA-LDA e PLS-DA para os modelos demonstraram que a classificação com os dados HSI-NIR foi alcançada com maior exatidão quando comparada aos modelos obtidos usando o NIR-convencional.

This thesis deals with modern statistical approaches and their application aimed at robust methods and neural network modelling. Selected methods are analyzed and applied on frequent practical problems in czech industry and technology. Topics and methods are to be benificial in real applications compared to currently used classical methods. Applicability and effectivity of the algorithms is verified and demonstrated on real studies and problems in czech industrial and research bodies. The great and unexploited potential of modern theoretical and computational capacity and the potential of new approaces to statistical modelling and methods. A significant result of this thesis is also an environment for software application development for data analysis with own programming language DARWin (Data Analysis Robot for Windows) for implemenation of effective numerical algorithms for extaction information from data. The thesis should be an incentive for boarder use of robust and computationally intensive methods as neural networks for modelling processes, quality control and generally better understanding of nature.

長期性資料(longitudinal data)的最主要特徵是為對多個被觀測個體在不同的時間點上重複測量一個或多個反應變數。而在分析長期性資料的方法中，Laird & Ware(1982)建議以線性混合效果模型(linear mixed effects model，LME)來進行估計分析，此模型方法中，資料可以允許遺失值，並可將受測個體間與個體內的變異分開說明。 另在配適最小平方法(OLS)的迴歸模型中，係數估計經常會受到異常值的影響，而Rousseeuw & Leroy(1987)提出最小消去平方法(least trimmed squares，LTS)的穩健迴歸模型，即是解決最小平方法中對於異常值敏感的問題。 本研究主要針對台灣股票預期報酬之三種模型：資本資產定價模型、特徵模型、因子模型分別以OLS、LTS、LME三種估計方法做配適，並比較配適模型之適當與否，樣本資料為民國七十年七月至九十年六月共252個月516家上市公司股票報酬。實證結果顯示，不論是採用OLS、LTS、LME的估計方法，股票報酬解釋變數：系統風險、公司規模、帳面權益對市值比、SMB、HML皆為股票報酬的顯著解釋因子；而在模型比較方面，不論是配適資本資產定價模型、特徵模型或因子模型，LME都較OLS為較適當配適模型。這顯示了在分析長期性資料時，LME的確是一個較佳的統計分析模型。

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。

本研究所使用違約機率（Probability of Default, 以下簡稱PD）的預測方法為Robust Logistic Regression（穩健羅吉斯迴歸），本研究發展且應用這個方法是基於下列兩個觀察：1. 極端值常常出現在橫剖面資料，而且對於實證結果往往有很大地影響，因而極端值必須要被謹慎處理。2. 當使用Logit Model（羅吉斯模型）估計違約率時，卻忽略極端值。試圖不讓資料中的極端值對估計結果產生重大的影響，進而提升預測的準確性，是本研究使用Logit Model並混合Robust Regression（穩健迴歸）的目的所在，而本研究是第一篇使用Robust Logistic Regression來進行PD預測的研究。 變數的選取上，本研究使用Z-SCORE模型中的變數，此外，在考慮公司的營收品質之下，亦針對公司的應收帳款週轉率而對相關變數做了調整。 本研究使用了一些信用風險模型效力驗證的方法來比較模型預測效力的優劣，本研究的實證結果為：針對樣本內資料，使用Robust Logistic Regression對於整個模型的預測效力的確有提升的效果；當營收品質成為模型變數的考量因素後，能讓模型有較高的預測效力。最後，本研究亦提出了一些重要的未來研究建議，以供後續的研究作為參考。 / The method implemented in PD calculation in this study is “Robust Logistic Regression”. We implement this method based on two reasons: 1. In panel data, outliers usually exist and they may seriously influence the empirical results. 2. In Logistic Model, outliers are not taken into consideration. The main purpose of implementing “Robust Logistic Regression” in this study is: eliminate the effects caused by the outliers in the data and improve the predictive ability. This study is the first study to implement “Robust Logistic Regression” in PD calculation. The same variables as those in Z-SCORE model are selected in this study. Furthermore, the quality of the revenue in a company is also considered. Therefore, we adjust the related variables with the company’s accounts receivable turnover ratio. Some validation methodologies for default risk models are used in this study. The empirical results of this study show that: In accordance with the in-sample data, implementing “Robust Logistic Regression” in PD calculation indeed improves the predictive ability. Besides, using the adjusted variables can also improve the predictive ability. In the end of this study, some important suggestions are given for the subsequent studies.