Nonlinear multivariate statistical techniques on fast computers offer the potential to capture more of the dynamics of the high dimensional, noisy systems underlying financial markets than traditional models, while making fewer restrictive assumptions. This thesis presents a collection of practical techniques to address important estimation and confidence issues for Radial Basis Function networks arising from such a data driven approach, including efficient methods for parameter estimation and pruning, a pointwise prediction error estimator, and a methodology for controlling the "data mining'' problem. Novel applications in the finance area are described, including customized, adaptive option pricing and stock price prediction.

<p>Recurrent neural networks have been used for time-series prediction with good results. In this dissertation recurrent neural networks are compared with time-delayed feed forward networks, feed forward networks and linear regression models on a prediction task. The data used in all experiments is real-world sales data containing two kinds of segments: campaign segments and non-campaign segments. The task is to make predictions of sales under campaigns. It is evaluated if more accurate predictions can be made when only using the campaign segments of the data.</p><p>Throughout the entire project a knowledge discovery process, identified in the literature has been used to give a structured work-process. The results show that the recurrent network is not better than the other evaluated algorithms, in fact, the time-delayed feed forward neural network showed to give the best predictions. The results also show that more accurate predictions could be made when only using information from campaign segments.</p>

Recurrent neural networks have been used for time-series prediction with good results. In this dissertation recurrent neural networks are compared with time-delayed feed forward networks, feed forward networks and linear regression models on a prediction task. The data used in all experiments is real-world sales data containing two kinds of segments: campaign segments and non-campaign segments. The task is to make predictions of sales under campaigns. It is evaluated if more accurate predictions can be made when only using the campaign segments of the data. Throughout the entire project a knowledge discovery process, identified in the literature has been used to give a structured work-process. The results show that the recurrent network is not better than the other evaluated algorithms, in fact, the time-delayed feed forward neural network showed to give the best predictions. The results also show that more accurate predictions could be made when only using information from campaign segments.

Finding significant relation between stock markets (including omnipresent volatility) and real economy of the US, Germany, Great Britain and Japan is the main aim of this thessis. If not found it is also the final conclusion. By means of time series analysis using artificial neural networks from the beginning of 2000 till the November of 2014 was proved that the strong single -- way relation between prime stock indices and GDP of chosen economies does exist. Highest quality of prediction was proved on the American and British economy. S&P 500, FTSE and VIX indicator made a precise prediction of future economic progress in the US and Great Britain for six to nine months ahead with 71% to 86% accuracy. The artificial neural networks proved an extraordinary ability to predict chosen financial time series regardless the actual position in a business cycle.

No description available.

Time series prediction plays a pivotal role in various areas, including for example finance, weather forecasting, and traffic analysis. In this study, time series of historical sales data from a packaging manufacturer is used to investigate the effects that clustering such data has on forecasting performance. An experiment is carried out in which the time series data is first clustered using two separate approaches: k-means and Self-Organizing Map (SOM). The clustering is feature-based, meaning that characteristics extracted from the time series are used to compute similarity, rather than the raw time series. Then, A set of Long Short-Term Memory models (LSTMs) are trained; one that is trained on the entire dataset (global model), separate models trained on each of the clusters (cluster-based models), and finally a number of models trained on individual time series that are proportionally sampled from the clusters (single models). By evaluating the LSTMs based on Mean Absolute Error (MAE) and Mean Squared Error (MSE), we assess their consistency and predictive potential. The results reveal a trade-off between the consistency and predictive performance of the models. The global LSTM model consistently exhibits more stable performance across all predictions, showcasing its ability to capture the overall patterns in the data. However, the cluster-based LSTM models demonstrate potential for improved predictive performance within specific clusters, albeit with higher variability. This suggests that certain clusters possess distinct characteristics that allow for better predictions within those subsets of the data. Finally, the single LSTM models trained on individual time series, showcase even wider spreads of scores. The analysis suggests that the availability of training data plays a crucial role in the robustness (i.e., the ability to consistently produce similar results) of the forecasting models, with the global model benefiting from a larger training set. The higher variability in performance seen for the models trained with smaller training sets indicates that certain time series may be easier or harder to predict. It seems that the noise that comes with a larger training set can be either beneficial or detrimental to the predictive performance of the forecasting model on any individual time series, depending on the characteristics of that particular sample. Further analysis is required to investigate the factors contributing to the varying performance within each cluster. Exploring feature scores associated with poorly performing clusters and identifying the key features that contribute to better performance in certain clusters could provide valuable insights. Understanding these factors might aid in developing tailored strategies for cluster-specific prediction tasks.

The goal of this thesis is to be able to predict customer traffic at postal service points. The expectation is that when customers are made aware of queue times at the service points, they will redistribute themselves to avoid standing in line. This boils down to a form of time series prediction problem. When working with time series prediction, there are potentially other factors that may help the models make a more accurate prediction. Factors that may affect people’s behavior are unlimited, but this thesis examines the effect of the external calendar variables (weekday, date and public holiday) and weather variables (temperature, precipitation and sun, among others) when making the predictions. Non-linear models are examined, with the focus on Multilayer Perceptrons (MLPs) and Long Short-Term Memory (LSTM) models that have shown promising results in time series prediction, and these models are referred to as Artificial Neural Networks (ANNs). Support Vector Regression (SVR), Autoregressive Moving Average (ARIMA) and statistical average models are used for comparison. The results show that using external variables as additional input to LSTM, MLP and SVR models increases the test prediction performance. Further, the MLP model generally performs better than the LSTM models. The results are acquired using six postal service points, and the final results are based on a six-fold cross validation across all six service points. The LSTM and MLP are able to better use the external variables and show greater adaptability during e.g. public holidays, compared with the SVR model. The ARIMA and historical average model show less accurate predictions compared with the aforementioned models. / Målet med detta examensarbete är att förutspå kundtrafik hos postserviceställen. Förhoppningen är att kunderna omfördelar sig själva om de får tillgång till kundtrafikprognoser för att undvika stå i kö. Detta resulterar i ett tidsserie-förutsägelseproblem. Vid sådana problem finns det potentiellt andra faktorer som kan påverka modellernas prediktioner positivt. Antalet faktorer som påverkar människors beteende är obegränsat, men detta examensarbete undersöker effekterna av att använda externa kalendervariabler (veckodag, datum och röd dag) och vädervariabler (temperatur, nederbörd och sol, bland annat). För att göra prediktionerna används främst de icke-linjära modellerna Multilayer Perceptron (MLP) och Long Short-Term Memory (LSTM), som båda refereras till som Artificial Neural Network (ANN). Båda modellerna har visat lovande resultat i liknande problem. Utöver dem används även modellerna Support Vector Regression (SVR) och Autoregressive Moving Average (ARIMA) samt det historiska genomsnittet som jämförelse. Resultaten visar på att om LSTM-, MLPoch SVR-modellerna får externa variabler som tilläggsinput så förbättras modellernas förutsägelser. Vidare presterar MLP-modellen generellt bättre än LSTMmodellen. Resultaten är skapade genom att använda sex stycken postserviceställen och de slutgiltiga resultaten är baserade på en 6-vägs korsvalidering för samtliga serviceställen. LSTMoch MLP-modellerna är bättre på att använda informationen från de externa variablerna och visar på större anpassningsförmåga, under till exempel röda dagar, jämfört med SVR-modellen. ARIMA-modellen och den historiska genomsnittsmodellen skapar sämre prediktioner än de förutnämndamodellerna.

In this thesis we examine methods for evaluating a traffic alarm system. Nuisance alarms can quickly increase the volume of alarms experienced by the alarm operator and obstruct their work. We propose two methods for removing a number of these nuisance alarms, so that events of higher priority can be targeted. A parallel correlation analysis demonstrated significant correlation between single and clusters of alarms, presenting a strong cause for causality. While a serial correlation was performed, it could not conclude evidence of consequential alarms. In order to assist Trafikverket with maintenance scheduling, a long short-term model (LSTM) model, to predict univariate time-series of discretely binned alarm sequences. Experiments conclude that the LSTM model provides higher precision for alarm sequences with higher repeatability and recurring patterns. For other, randomly occurring alarms, the model performs unsatisfactory. / Den här examensuppsatsen granskar olika metoder för att utvärdera ett larmsystem med inriktning mot trafiksäkerhet. Störande larm kan skapa stora mängder larm som försvårar arbetet för larmoperatörer. Vi föreslår två metoder för att avlägsna störande larm, så att uppmärksamhet kan riktas mot varningar med högre prioritet. En parallell korrelationsanalys som demonstrerade hög korrelation mellan både enskilda och kluster av larm. Detta presenterar ett starkt orsakssamband. En korskorrelation utfördes även, men denna kunde inte fastställa existens av s.k. följdlarm. För att assistera Trafikverket med schemaläggning av underhåll har en long short-term memory (LSTM) modell implementerats för att förutspå univariata tidsserier av diskretiserade larmsekvenser. Utförda experiment sammanfattar att LSTM modellen presterar bättre för larmsekvenser med återkommande mönster. För mera slumpmässigt genererade larmsekvenser, presterar modellen med lägre precision.

No description available.

Made available in DSpace on 2017-07-10T17:11:50Z (GMT). No. of bitstreams: 1 JORGE AIKES JUNIOR.PDF: 2050278 bytes, checksum: f5bae18bbcb7465240488c45b2c813e7 (MD5) Previous issue date: 2012-04-11 / Time series can be understood as any set of observations which are time ordered. Among the many possible tasks appliable to temporal data, one that has attracted increasing interest, due to its various applications, is the time series forecasting. The k-Nearest Neighbor - Time Series Prediction (kNN-TSP) algorithm is a non-parametric method for forecasting time series. One of its advantages, is its easiness application when compared to parametric methods. Even though its easier to define kNN-TSP s parameters, some issues remain opened. This research is focused on the study of one of these parameters: the similarity measure. This parameter was empirically evaluated using various similarity measures in a large set of time series, including artificial series with seasonal and chaotic characteristics, and several real world time series. It was also carried out a case study comparing the predictive accuracy of the kNN-TSP algorithm with the Moving Average (MA), univariate Seasonal Auto-Regressive Integrated Moving Average (SARIMA) and multivariate SARIMA methods in a time series of a Korean s hospital daily patients flow in the Emergency Department. This work also proposes an approach to the development of a hybrid similarity measure which combines characteristics from several measures. The research s result demonstrated that the Lp Norm s measures have an advantage over other measures evaluated, due to its lower computational cost and for providing, in general, greater accuracy in temporal data forecasting using the kNN-TSP algorithm. Although the literature in general adopts the Euclidean similarity measure to calculate de similarity between time series, the Manhattan s distance can be considered an interesting candidate for defining similarity, due to the absence of statistical significant difference and to its lower computational cost when compared to the Euclidian measure. The measure proposed in this work does not show significant results, but it is promising for further research. Regarding the case study, the kNN-TSP algorithm with only the similarity measure parameter optimized achieves a considerably lower error than the MA s best configuration, and a slightly greater error than the univariate e multivariate SARIMA s optimal settings presenting less than one percent of difference. / Séries temporais podem ser entendidas como qualquer conjunto de observações que se encontram ordenadas no tempo. Dentre as várias tarefas possíveis com dados temporais, uma que tem atraído crescente interesse, devido a suas várias aplicações, é a previsão de séries temporais. O algoritmo k-Nearest Neighbor - Time Series Prediction (kNN-TSP) é um método não-paramétrico de previsão de séries temporais que apresenta como uma de suas vantagens a facilidade de aplicação, quando comparado aos métodos paramétricos. Apesar da maior facilidade na determinação de seus parâmetros, algumas questões relacionadas continuam em aberto. Este trabalho está focado no estudo de um desses parâmetros: a medida de similaridade. Esse parâmetro foi avaliado empiricamente utilizando diversas medidas de similaridade em um grande conjunto de séries temporais que incluem séries artificiais, com características sazonais e caóticas, e várias séries reais. Foi realizado também um estudo de caso comparativo entre a precisão da previsão do algoritmo kNN-TSP e a dos métodos de Médias Móveis (MA), Auto-regressivos de Médias Móveis Integrados Sazonais (SARIMA) univariado e SARIMA multivariado, em uma série de fluxo diário de pacientes na Área de Emergência de um hospital coreano. Neste trabalho é ainda proposta uma abordagem para o desenvolvimento de uma medida de similaridade híbrida, que combine características de várias medidas. Os resultados obtidos neste trabalho demonstram que as medidas da Norma Lp apresentam vantagem sobre as demais medidas avaliadas, devido ao seu menor custo computacional e por apresentar, em geral, maior precisão na previsão de dados temporais utilizando o algoritmo kNN-TSP. Apesar de na literatura, em geral, a medida Euclidiana ser adotada como medida de similaridade, a medida Manhattan pode ser considerada candidata interessante para definir a similaridade entre séries temporais, devido a não apresentar diferença estatisticamente significativa com a medida Euclidiana e possuir menor custo computacional. A medida proposta neste trabalho, não apresenta resultados significantes, mas apresenta-se promissora para novas pesquisas. Com relação ao estudo de caso, o algoritmo kNN-TSP, com apenas o parâmetro de medida de similaridade otimizado, alcança um erro consideravelmente inferior a melhor configuração com MA, e pouco maior que as melhores configurações dos métodos SARIMA univariado e SARIMA multivariado, sendo essa diferença inferior a um por cento.