Physical Sciences and Mathematics Commons^™

Reevaluating Texas Energy Market Forecasts In The Wake Of Recent Extreme Weather Events, Robert A. Derner, Richard W. Butler Ii, Alexandria Neff, Adam R. Ruthford

SMU Data Science Review

This paper provides updated forecasts of energy demand in Texas and recognizes the impact of sustainable energy. It is important that the forecasts of the adoption of sustainable energy are reexamined after Winter Storm Uri crippled the Texas power grid and left many without power. This storm highlighted the issues the Texas power grid had and has continued to struggle with in supplying the state with energy. This paper will offer an overview of the relevant literature on the adoption of sustainable energy and relevant events that have occurred in the state of Texas that will give the reader the …

Traditional Vs Machine Learning Approaches: A Comparison Of Time Series Modeling Methods, Miguel E. Bonilla Jr., Jason Mcdonald, Tamas Toth, Bivin Sadler

SMU Data Science Review

In recent years, various new Machine Learning and Deep Learning algorithms have been introduced, claiming to offer better performance than traditional statistical approaches when forecasting time series. Studies seeking evidence to support the usage of ML/DL over statistical approaches have been limited to comparing the forecasting performance of univariate, linear time series data. This research compares the performance of traditional statistical-based and ML/DL methods for forecasting multivariate and nonlinear time series.

A State Space Modeling Approach To Eeg Artifact Removal, Patrick B. Rafael

Graduate Research Theses & Dissertations

In this work, a state space modeling approach is applied to an Electroencephalography(EEG) recording for the purpose of artifact removal, and is compared against Independent Components Analysis (ICA), the current gold standard. Issues of model identifiability are touched on, and Hamiltonian Monte Carlo (HMC) is used to estimate a linear non-Gaussian state space model. Results show that estimating such a model is a nontrivial matter, and the full utility of the state space approach remains to be demonstrated.

Predicting Insulin Pump Therapy Settings, Riccardo L. Ferraro, David Grijalva, Alex Trahan

SMU Data Science Review

Millions of people live with diabetes worldwide [7]. To mitigate some of the many symptoms associated with diabetes, an estimated 350,000 people in the United States rely on insulin pumps [17]. For many of these people, how effectively their insulin pump performs is the difference between sleeping through the night and a life threatening emergency treatment at a hospital. Three programmed insulin pump therapy settings governing effective insulin pump function are: Basal Rate (BR), Insulin Sensitivity Factor (ISF), and Carbohydrate Ratio (ICR). For many people using insulin pumps, these therapy settings are often not correct, given their physiological needs. While …

Penalized Estimation Of Autocorrelation, Xiyan Tan

All Dissertations

This dissertation explored the idea of penalized method in estimating the autocorrelation (ACF) and partial autocorrelation (PACF) in order to solve the problem that the sample (partial) autocorrelation underestimates the magnitude of (partial) autocorrelation in stationary time series. Although finite sample bias corrections can be found under specific assumed models, no general formulae are available. We introduce a novel penalized M-estimator for (partial) autocorrelation, with the penalty pushing the estimator toward a target selected from the data. This both encapsulates and differs from previous attempts at penalized estimation for autocorrelation, which shrink the estimator toward the target value of zero. …

Statistical Modeling Of Daily Confirmed Covid-19 Cases And Deaths In Europe And United States, Zerui Zhang

Master's Theses (2009 -)

A novel coronavirus disease was first discovered in Wuhan, China, in December 2019. This new coronavirus named COVID-19 has rapidly spread and become a global threat affecting almost all the countries in the world. Therefore, it is important to know the trend of coronavirus disease to mitigate its effects. A good prediction model is crucial for the health care system to understand the trend of the COVID-19. This study aims to construct a good prediction model. Firstly, we detect change points of the time series data of daily confirmed cases and deaths of COVID-19 in the United States and Europe, …

Machine Learning Based Restaurant Sales Forecasting, Austin B. Schmidt

University of New Orleans Theses and Dissertations

To encourage proper employee scheduling for managing crew load, restaurants have a need for accurate sales forecasting. We predict partitions of sales days, so each day is broken up into three sales periods: 10:00 AM-1:59 PM, 2:00 PM-5:59 PM, and 6:00 PM-10:00 PM. This study focuses on the middle timeslot, where sales forecasts should extend for one week. We gather three years of sales between 2016-2019 from a local restaurant, to generate a new dataset for researching sales forecasting methods.

Outlined are methodologies used when going from raw data to a workable dataset. We test many machine learning models on …

Modeling Time Series With Conditional Heteroscedastic Structure, Ratnayake Mudiyanselage Isuru Panduka Ratnayake

Doctoral Dissertations

"Models with a conditional heteroscedastic variance structure play a vital role in many applications, including modeling financial volatility. In this dissertation several existing formulations, motivated by the Generalized Autoregressive Conditional Heteroscedastic model, are further generalized to provide more effective modeling of price range data well as count data. First, the Conditional Autoregressive Range (CARR) model is generalized by introducing a composite range-based multiplicative component formulation named the Composite CARR model. This formulation enables a more effective modeling of the long and short-term volatility components present in price range data. It treats the long-term volatility as a stochastic component that in …

Demand Forecasting In Wholesale Alcohol Distribution: An Ensemble Approach, Tanvi Arora, Rajat Chandna, Stacy Conant, Bivin Sadler, Robert Slater

SMU Data Science Review

In this paper, historical data from a wholesale alcoholic beverage distributor was used to forecast sales demand. Demand forecasting is a vital part of the sale and distribution of many goods. Accurate forecasting can be used to optimize inventory, improve cash ow, and enhance customer service. However, demand forecasting is a challenging task due to the many unknowns that can impact sales, such as the weather and the state of the economy. While many studies focus effort on modeling consumer demand and endpoint retail sales, this study focused on demand forecasting from the distributor perspective. An ensemble approach was applied …

Demand Forecasting For Alcoholic Beverage Distribution, Lei Jiang, Kristen M. Rollins, Meredith Ludlow, Bivin Sadler

SMU Data Science Review

Forecasting demand is one of the biggest challenges in any business, and the ability to make such predictions is an invaluable resource to a company. While difficult, predicting demand for products should be increasingly accessible due to the volume of data collected in businesses and the continuing advancements of machine learning models. This paper presents forecasting models for two vodka products for an alcoholic beverage distributing company located in the United States with the purpose of improving the company’s ability to forecast demand for those products. The results contain exploratory data analysis to determine the most important variables impacting demand, …

Home Sales As A Time Series Model, Noah R. Hellenthal

Williams Honors College, Honors Research Projects

Rational Expectations Hypothesis is an economic theorem that states that our best way to predict the future is by looking at the past. While this theory is typically used to address inflation, the same concept can be used when predicting future home sales. With the failure of subprime mortgages and the burst of the housing market bubble in 2008, home sales are proven to be an appropriate indication of how the U.S. economy is performing. Through time series analysis, I will be able to construct a model with monthly home sales data from the U.S. Census Bureau. Due to seasonality …

Applications Of Dynamic Linear Models To Random Allocation Models, Albert H. Lee Iii

Theses and Dissertations

Although advances in modern computational algorithms have provided researchers the ability to work problems which were once too computationally complex to solve, problems with high computation or large parameter spaces still remain. Problems such as those involving Time Series can be such problems. Chapter 1 looks at the the use of Exponentially Weighted Moving Averages developed by \citep{holt2004forecasting, winters1960forecasting} which were thought to provide sufficient solutions to these Time Series. A discussion is provided which illustrates the shortcomings of the EWMA and how its infinite number of possible starting values provides the modeler with an endless number of possible solutions …

A Multi-Step Approach To Modeling The 24-Hour Daily Profiles Of Electricity Load Using Daily Splines, Abdelmonaem Jornaz, V. A. Samaranayake

Mathematics and Statistics Faculty Research & Creative Works

Forecasting of real-time electricity load has been an important research topic over many years. Electricity load is driven by many factors, including economic conditions and weather. Furthermore, the demand for electricity varies with time, with different hours of the day and different days of the week having an effect on the load. This paper proposes a hybrid load-forecasting method that combines classical time series formulations with cubic splines to model electricity load. It is shown that this approach produces a model capable of making short-term forecasts with reasonable accuracy. In contrast to forecasting models that utilize a multitude of regressor …

A Comparison Of The Predictive Ability Of Logistic Regression And Time Series Analysis On Business Credit Data, Lauren Staples

Published and Grey Literature from PhD Candidates

The credit industry creates models to determine the risk of lending money to consumers as well as to commercial customers. These models are heavily regulated in the U.S. as well as in other countries. Model inputs must be explainable to customers as well as to regulators. Two such modeling approaches that are currently commonly used are logistic regression models and time series models. This paper steps through the preprocessing and model building of these two models on a large commercial data set and compares the predictive ability of these two methods. The two models achieved similar accuracy results: the logistic …

Non-Stationary Counts With Mixture Distributions, Ziqiang Lin

Legacy Theses & Dissertations (2009 - 2024)

We study a new non--stationary mixture Pengram and thinning model for time series of counts that include the effect of covariate variables on the outcome variable. Properties of the model and performance are discussed. It has a simpler likelihood function than the non--stationary INAR(1) model and therefore MLE estimators for the model's parameters are easier to find. Therefore the model offers an alternative to non--stationary INAR(1).

Spatio-Temporal Frequency Separation With Application Of Kolmogorov-Zurbenko Filters To The Multivariate Analysis Of Melanoma Prevalence, Edward Valachovic

Legacy Theses & Dissertations (2009 - 2024)

Time Series Analysis is the observation of variables recorded across time. Observations are visualized and analysis often performed in the native time domain. It is common for a time series to be the dependent variable of more than one factor. Several factors can have concurrent and combined effects. The time domain presents an obstacle due to constructive and destructive interference of factors at each time point. Unless effects are clearly pronounced and separable, the entanglement of factors along with the presence and intensity of random variation can obscure true relationships.

Statistical Methods On Risk Management Of Extreme Events, Zijing Zhang

Doctoral Dissertations

The goal of the dissertation is the investigation of financial risk analysis methodologies, using the schemes for extreme value modeling as well as techniques from copula modeling. Extreme value theory is concerned with probabilistic and statistical questions re- lated to unusual behavior or rare events. The subject has a rich mathematical theory and also a long tradition of applications in a variety of areas. We are interested in its application in risk management, with a focus on estimating and forcasting the Value-at-Risk of financial time series data. Extremal data are inherently scarce, thus making inference challenging. In order to obtain …

Garch(1,1) With Sifted Gamma-Distributed Errors, Alan C. Budd

Electronic Theses and Dissertations

Typical General Autoregressive Conditional Heteroskedastic (GARCH) processes involve normally-distributed errors, and they model strictly-positive error processes poorly. This thesis will present a method for estimating the parameters of a GARCH(1,1) process with shifted Gamma-distributed errors, conduct a simulation study to test the method, and apply the method to real time series data.

Sensitivity Of Mixed Models To Computational Algorithms Of Time Series Data, Gunaime Nevine

Doctoral Dissertations

Statistical analysis is influenced by implementation of the algorithms used to execute the computations associated with various statistical techniques. Over many years; very important criteria for model comparison has been studied and examined, and two algorithms on a single dataset have been performed numerous times. The goal of this research is not comparing two or more models on one dataset, but comparing models with numerical algorithms that have been used to solve them on the same dataset.

In this research, different models have been broadly applied in modeling and their contrasting which are affected by the numerical algorithms in different …

The Bootstrap Estimation In Time Series, Yun Liu

Dissertations, Master's Theses and Master's Reports

Time series, a special case in dependent data sequence, is widely used in many fields. In time series, linear process models are quite popularly used. General form of linear process indicates the time dependence property of time series, AR(p), MA(q) and ARMA(p;,q) models are all linear process models. In this report, simulations are based on the simplest models of these linear process models, such as AR(1), MA(1) and ARMA(1,1) models. AR(1)-SEASON, which is developed based on AR(1) model by changing the weight of residuals, is also considered in this report. To deal with dependent data sequence, common methods which aim …

Statistical Applications In Wildfire Management And Prediction, Lengyi Han

Electronic Thesis and Dissertation Repository

This thesis develops statistical methods and models and applies them
to problems related to forest fires. The unifying goal of the work is to provide a data analytic basis for quantifying the uncertainty surrounding fire ignition and fire growth which builds on existing theory where possible.

The main body of the thesis is comprised of three research papers. The Fire Weather Index (FWI) plays an important role in fire management and is central to the first two papers. In the first instance, the block bootstrap confidence interval method is used to deal nonparametrically with the dependence in the FWI data. …

High Frequency Data: Modeling Durations Via The Acd And Log Acd Models, Lilian Cheung

Honors Scholar Theses

This thesis proposes a method of finding initial parameter estimates in the Log ACD₁ model for use in recursive estimation. The recursive estimating equations method is applied to the Log ACD₁ model to find recursive estimates for the unknown parameters in the model. A literature review is provided on the ACD and Log ACD models, and on the theory of estimating equations. Monte Carlo simulations indicate that the proposed method of finding initial parameter estimates is viable. The parameter estimation process is demonstrated by fitting an ACD model and a Log ACD model to a set of IBM …

The Modified R A Robust Measure Of Association For Time Series, Muhammad Irfan Malik

irfan.phdet24@iiu.edu.pk

Since times of Yule (1926), it is known that correlation between two time series can produce spurious results. Granger and Newbold (1974) see the roots of spurious correlation in non-stationarity of the time series. However the study of Granger, Hyung and Jeon (2001) prove that spurious correlation also exists in stationary time series. These facts make the correlation coefficient an unreliable measure of association. This paper proposes ‘Modified R’ as an alternate measure of association for the time series. The Modified R is robust to the type of stationarity and type of deterministic part in the time series. The performance …

Repeat Sales House Price Index Methodology, Chaitra Nagaraja, Lawrence Brown, Susan Wachter

Chaitra H Nagaraja

No abstract provided.

Constructing And Evaluating An Autoregressive House Price Index, Chaitra Nagaraja, Lawrence Brown

Chaitra H Nagaraja

No abstract provided.

Decomposition Of High Frequency Data In Components: Visualization And Interpretative Models, Carlo Drago

Carlo Drago

No abstract provided.

An Autoregressive Approach To House Price Modeling, Chaitra Nagaraja, Lawrence Brown, Linda Zhao

Chaitra H Nagaraja

No abstract provided.

Methods For The Estimation Of Missing Values In Time Series, David S. Fung

Theses: Doctorates and Masters

Time Series is a sequential set of data measured over time. Examples of time series arise in a variety of areas, ranging from engineering to economics. The analysis of time series data constitutes an important area of statistics. Since, the data are records taken through time, missing observations in time series data are very common. This occurs because an observation may not be made at a particular time owing to faulty equipment, lost records, or a mistake, which cannot be rectified until later. When one or more observations are missing it may be necessary to estimate the model and also …