Forecasting economic recessions has been a major topic of interest for economists, decisionmakers and the public alike. In this study we ventured to analyse how changes in economic output or GDP is related to changes in consumer sentiment and house prices in the Swedish economy. For the purpose we employed simple vector autoregression (VAR) methodology that has the benefit of modelling dynamic relationships across time. Our results indicated that consumer sentiment, measured by Consumer Confidence Index, and housing prices, measured by the Housing Index, are indeed related to GDP meaning that these variables can be used for forecasting GDP and potentially be indicative of imminent recessions. Overall, this study contributes to the body of knowledge on economic forecasting and VAR methodologies.

The purpose of this research is to simulate warm-season mesoscale convective systems (MCSs) using the Weather Research and Forecasting Model (WRF) to determine whether modeled atmospheric variables are capable of discriminating between derecho formation and intensity. Fifty total events are selected with half being derecho-producing MCSs and half being non-derecho producing MCSs. WRF is used to model each event with a high-resolution domain centered over the Midwest using the North American Regional Reanalysis (NARR) dataset as initial and boundary conditions. Atmospheric conditions downstream of the MCS damage path are compared to thresholds established by previous research to determine if the model accurately simulates the expected environment. The goal of the research is to gain insight into how well a high-resolution model can simulate the environment that is expected. It is anticipated that the model will be able to distinguish between environments associated with a derecho-producing MCS and a non-derecho MCS.

No description available.

No description available.

No description available.

As the world becomes more urbanized and heavy precipitation events increase in frequency and intensity, urban flooding is an emerging concern. Urban flooding is caused when heavy rainfall collects on the landscape, exceeding the capacity of drainage systems to effectively convey runoff. Unlike riverine and coastal flooding, urban flooding occurs frequently, and its risks and impacts are not restricted to areas within floodplains or near bodies of water. The objective of this dissertation is to improve our understanding of urban flooding and our capability to predict it through the development of tools and knowledge to assist with its analysis, modeling, and forecasting. To do this, three research objectives were fulfilled. First, the Stream Hydrology And Rainfall Knowledge System (SHARKS) app was developed to improve upon existing real-time hydrologic and meteorological data retrieval/visualization platforms through the integration of analysis tools to study the hydrologic processes influencing urban flooding. Next, the ability to simulate the hydrologic response of urban watersheds with large storm sewer networks was compared between the fully distributed Gridded Surface/Subsurface Hydrologic Analysis (GSSHA) model and the semi-distributed Storm Water Management Model (SWMM). Finally, the Probabilistic Urban Flash Flood Information Nexus (PUFFIN) application was created to help users evaluate the probability of urban flash flooding and to identify specific infrastructure components at risk through the integration of high-resolution quantitative precipitation forecasting, ensemble forecasting, and hydrologic and hydraulic modeling. The outcomes of this dissertation provide municipalities with tools and knowledge to assist them throughout the process of developing solutions to their site-specific urban flooding issues. Specifically, tools are provided to rapidly analyze and respond to rainfall and streamflow/depth information during intense rain events and to perform retrospective analysis of long-term hydrological processes. Evaluations are included to help guide the selection of hydrologic and hydraulic models for modeling urban flooding, and a new proactive paradigm of probabilistic flash flood guidance for urban areas is introduced. Finally, several potential directions for future work are recommended. / Doctor of Philosophy / As the world becomes more urbanized and heavy precipitation events increase in frequency and intensity, urban flooding is an emerging concern. Urban flooding is caused when heavy rainfall collects on the landscape, exceeding the capacity of drainage systems to effectively convey runoff. Unlike riverine and coastal flooding, urban flooding occurs frequently, and its risks and impacts are not restricted to areas within floodplains or near bodies of water. The objective of this dissertation is to improve our understanding of urban flooding and our capability to predict it through the development of tools and knowledge to assist with its analysis, modeling, and forecasting. To do this, three research objectives were fulfilled. First, the Stream Hydrology And Rainfall Knowledge System (SHARKS) app was developed to improve upon existing real-time hydrologic and meteorological data retrieval/visualization platforms through the integration of analysis tools to study the hydrologic processes influencing urban flooding. Next, the ability to simulate the hydrologic response of urban watersheds with large storm sewer networks was compared between the fully distributed Gridded Surface/Subsurface Hydrologic Analysis (GSSHA) model and the semi-distributed Storm Water Management Model (SWMM). Finally, the Probabilistic Urban Flash Flood Information Nexus (PUFFIN) application was created to help users evaluate the probability of urban flash flooding and to identify specific infrastructure components at risk through the integration of high-resolution quantitative precipitation forecasting, ensemble forecasting, and hydrologic and hydraulic modeling. The outcomes of this dissertation provide municipalities with tools and knowledge to assist them throughout the process of developing solutions to their site-specific urban flooding issues. Specifically, tools are provided to rapidly analyze and respond to rainfall and streamflow/depth information during intense rain events and to perform retrospective analysis of long-term hydrological processes. Evaluations are included to help guide the selection of hydrologic and hydraulic models for modeling urban flooding, and a new proactive paradigm of probabilistic flash flood guidance for urban areas is introduced. Finally, several potential directions for future work are recommended.

Over the past two decades, a surge of interest in the area of forecasting has produced a number of statistical models available for predicting the winners of U.S. presidential elections. While historically the domain of individuals outside the scholarly community - such as political strategists, pollsters, and journalists - presidential election forecasting has become increasingly mainstream, as a number of prominent political scientists entered the forecasting arena. With the goal of making accurate predictions well in advance of the November election, these forecasters examine several important election "fundamentals" previously shown to impact national election outcomes. In general, most models employ some measure of presidential popularity as well as a variety of indicators assessing the economic conditions prior to the election. Advancing beyond the traditional, non-scientific approaches employed by prognosticators, politicos, and pundits, today's scientific models rely on decades of voting behavior research and sophisticated statistical techniques in making accurate point estimates of the incumbent's or his party's percentage of the popular two-party vote. As the latest evolution in presidential forecasting, these models represent the most accurate and reliable method of predicting elections to date. This thesis provides an assessment of forecasting models' underlying epistemological assumptions, theoretical foundations, and methodological approaches. Additionally, this study addresses forecasting's implications for related bodies of literature, particularly its impact on studies of campaign effects. / Master of Arts

One of the most challenging topics of deep learning (DL) is the analysis of temporal series in complex real-world scenarios. The majority of proposed DL methods tend to simplify such environments without considering several factors. The first part of this thesis focuses on developing video surveillance and sports analytic systems, in which obstacles, social interactions, and flow directions are relevant aspects. A DL model is then proposed to predict future paths, taking into account human interactions sharing a common memory, and favouring the most common paths through belief maps. Another model is proposed, adding the possibility to consider agents' goals. This aspect is particularly relevant in sports games where players can share objectives and tactics. Both the proposed models rely on the common hypothesis that the whole amount of labelled data is available from the beginning of the analysis, without evolving. This can be a strong simplification for most real-world scenarios, where data is available as a stream and changes over time. Thus, a theoretical model for continual learning is then developed to face problems where few data come as a stream, and labelling them is a hard task. Finally, continual learning strategies are applied to one of the most challenging scenarios for DL: financial market predictions. A collection of state-of-the-art continual learning techniques are applied to financial indicators representing temporal data. Results achieved during this PhD show how artificial intelligence algorithms can help to solve real-world problems in complex and time-evolving scenarios.

This thesis studies three important topics in modeling financial volatility. First, the jump clustering in ex post variance and its implications on forecasting, second, the underlying distribution of stochastic volatility and third, the role of non-Gaussian multivariate return distribution combined with a realized GARCH framework. The first chapter is on variance jumps. Financial markets present unexpected and large jumps, due to unobserved news flow. I focus on modeling the ex post variance jumps, their time- dependent arrivals and their sizes. I use a discrete-time bivariate model, with two autoregressive components which capture the long and short-run memory of the ex post variance measures. I estimate contemporaneous and time-dependent jumps in the log-measures of realized variance and bipower variation. The results from S&P500 show that the variance jumps are frequent and persistent. I examine the ability of jumps to forecast returns and ex post variance densities over horizons of up to 50 days out-of-sample. Modeling jumps significantly improves ex post variance density forecasts for all horizons and improves forecasts of the returns density. In the second chapter I explore the empirical non-Gaussian features of stochastic volatility. The standard assumption in a stochastic volatility specification is typically a restrictive Gaussian AR(1) structure. I drop this assumption and instead I assume that latent log-volatility follows an infinite mixture of normals with a Dirichlet process prior. The ex post measure of realized variance is used as a source of information to help identify the unknown distribution of log- volatility. Results from major stock indices show strong evidence of non-Gaussian distributional behaviour of volatility. The proposed framework captures asymmetry and thick tails in returns as well as realized variance. In out-of-sample forecasting, the new model provides improved density forecasts for returns, negative returns and log-realized variance. In the third chapter a new approach for multivariate realized GARCH models is proposed. Two new extensions that have non-Gaussian innovations are developed. The first one is a parametric version, with multivariate-t innovations. The second one is a nonparametric approximation of the return distribution using an infinite mixture of multivariate normals given a Dirichlet process prior. The proposed models are based on the assumption that the realized covariance follows an Inverse Wishart distribution with conditional mean set to the conditional covariance of returns. The benefits of the proposed models are demonstrated from density forecasting and portfolio applications. Results from two equity datasets indicate that modeling the tail behaviour improves return density forecasting compared to the Gaussian assumption. The proposed models produce the least volatile global minimum variance portfolios out-of-sample and provide improved forecasts of Value-at-Risk and Expected Shortfall. / Thesis / Doctor of Business Administration (DBA)

This research designs a purchasing decision support system (PDSS) to assist real-world decision makings on whether to purchase or to sub-hire for equipment shortfalls problem, and to avoid shortage loss for rental business. Research methodology includes an extensive literature review on decision support systems, rental industry, and forecasting methods. A case study was conducted in a rental company to learn the real world problem and to develop the research topics. A data converter is developed to recover the missing data and transform data sets to the accumulative usage data for the forecasting model. Simulations on a number of forecasting methods was carried out to select the best method for the research data based on the lowest forecasting errors. A hybrid forecasting approach is proposed by adding company revenue data as a parameter, in addition to the selected regression model to further reduce the forecasting error. Using the forecasted equipment usage, a two stage PDSS model was constructed and integrated to the forecasting model and data converter. This research fills the gap between decision support system and rental industry. The PDSS now assists the rental company on equipments buy or hire decisions. A hybrid forecasting method has been introduced to improve the forecasting accuracy significantly. A dada converter is designed to efficiently resolve data missing and data format problems, which is very common in real world.