Physical Sciences and Mathematics Commons^™

Temporal Implicit Multimodal Networks For Investment And Risk Management, Meng Kiat Gary Ang, Ee-Peng Lim

Research Collection School Of Computing and Information Systems

Many deep learning works on financial time-series forecasting focus on predicting future prices/returns of individual assets with numerical price-related information for trading, and hence propose models designed for univariate, single-task, and/or unimodal settings. Forecasting for investment and risk management involves multiple tasks in multivariate settings: forecasts of expected returns and risks of assets in portfolios, and correlations between these assets. As different sources/types of time-series influence future returns, risks, and correlations of assets in different ways, it is also important to capture time-series from different modalities. Hence, this article addresses financial time-series forecasting for investment and risk management in a …

In Pursuit Of Consumption-Based Forecasting, Charles Chase, Kenneth B. Kahn

Marketing Faculty Publications

[Introduction] Today's most mature, most sophisticated, best-in-class forecasting is what we call consumption-based forecasting (CBF). In contrast, the least sophisticated companies typically do not forecast at all, but rather set financial targets based on management expectations. Companies beginning to use statistical forecasting techniques usually take a supply-centric orientation, relying on time series techniques applied to shipment and/or order history. The next stage of progression is to incorporate promotions data, economic data, and market data alongside supply-centric data so that regression and other advanced analytics can be used. Companies pursing CBF utilize even more advanced capabilities to capture, examine, and understand …

Physics-Informed Deep Learning With Kalman Filter Mixture For Traffic State Prediction, Niharika Deshpande, Hyoshin (John) Park

Engineering Management & Systems Engineering Faculty Publications

Accurate traffic forecasting is crucial for understanding and managing congestion for efficient transportation planning. However, conventional approaches often neglect epistemic uncertainty, which arises from incomplete knowledge across different spatiotemporal scales. This study addresses this challenge by introducing a novel methodology to establish dynamic spatiotemporal correlations that captures the unobserved heterogeneity in travel time through distinct peaks in probability density functions, guided by physics-based principles. We propose an innovative approach to modifying both prediction and correction steps of the Kalman Filter (KF) algorithm by leveraging established spatiotemporal correlations. Central to our approach is the development of a novel deep learning model …

Lightning Forecast From Chaotic And Incomplete Time Series Using Wavelet De-Noising And Spatiotemporal Kriging, Jared K. Nystrom, Raymond Hill, Andrew J. Geyer, Joseph J. Pignatiello Jr., Eric Chicken

Faculty Publications

Purpose: Present a method to impute missing data from a chaotic time series, in this case lightning prediction data, and then use that completed dataset to create lightning prediction forecasts.

Design/Methodology/Approach: Using the technique of spatiotemporal kriging to estimate data that is autocorrelated but in space and time. Using the estimated data in an imputation methodology completes a dataset used in lighting prediction.

Findings: The techniques provided prove robust to the chaotic nature of the data, and the resulting time series displays evidence of smoothing while also preserving the signal of interest for lightning prediction.

Abstract © Emerald Publishing …

Learning Deep Time-Index Models For Time Series Forecasting, Jiale Gerald Woo, Chenghao Liu, Doyen Sahoo, Akshat Kumar, Steven Hoi

Research Collection School Of Computing and Information Systems

Deep learning has been actively applied to time series forecasting, leading to a deluge of new methods, belonging to the class of historicalvalue models. Yet, despite the attractive properties of time-index models, such as being able to model the continuous nature of underlying time series dynamics, little attention has been given to them. Indeed, while naive deep timeindex models are far more expressive than the manually predefined function representations of classical time-index models, they are inadequate for forecasting, being unable to generalize to unseen time steps due to the lack of inductive bias. In this paper, we propose DeepTime, a …

Hyper-Local Weather Predictions With The Enhanced General Urban Area Microclimate Predictions Tool, Kevin A. Adkins, William Becker, Sricharan Ayyalasomayajula, Steven Lavenstein, Kleoniki Vlachou, David Miller, Marc Compere, Avinash Muthu Krishnan, Nickolas Macchiarella

Publications

This paper presents enhancements to, and the demonstration of, the General Urban area Microclimate Predictions tool (GUMP), which is designed to provide hyper-local weather predictions by combining machine-learning (ML) models and computational fluid dynamic (CFD) simulations. For the further development and demonstration of GUMP, the Embry–Riddle Aeronautical University (ERAU) campus was used as a test environment. Local weather sensors provided data to train ML models, and CFD models of urban- and suburban-like areas of ERAU’s campus were created and iterated through with a wide assortment of inlet wind speed and direction combinations. ML weather sensor predictions were combined with best-fit …

Time Series Analysis Of Major League Baseball Organizations’ Fan Attendance, Joseph Molis

Honors Program Theses and Projects

Throughout baseball’s rich and long history, fans have been one of the most integral parts of the game. However, in recent years, baseball has seen a decrease in fans, allegedly due to the pace of play, or the length of games. Baseball games can take up to four hours to complete, and in today’s fast-moving society where all information is at one’s fingertips, it is believed that baseball’s slower pace turns people away from the game. However, how true is that? The primary goal of this project is to build models to accurately forecast fan attendance for every Major League …

The Effect Of Advection On The Three Dimensional Distribution Of Turbulent Kinetic Energy And Its Generation In Idealized Tropical Cyclone Simulations, Joshua B. Wadler, David S. Nolan, Jun A. Zhang, Lynn K. Shay, Joseph B. Olsen, Joseph J. Cione

Publications

The distribution of turbulent kinetic energy (TKE) and its budget terms is estimated in simulated tropical cyclones (TCs) of various intensities. Each simulated TC is subject to storm motion, wind shear, and oceanic coupling. Different storm intensities are achieved through different ocean profiles in the model initialization. For each oceanic profile, the atmospheric simulations are performed with and without TKE advection. In all simulations, the TKE is maximized at low levels (i.e., below 1 km) and ∼0.5 km radially inward of the azimuthal-mean radius of maximum wind speed at 1-km height. As in a previous study, the axisymmetric TKE decreases …

Uncertaintyfusenet: Robust Uncertainty-Aware Hierarchical Feature Fusion Model With Ensemble Monte Carlo Dropout For Covid-19 Detection, Moloud Abdar, Soorena Salari, Sina Qahremani, Hak-Keung Lam, Fakhreddine (Fakhri) Karray, Sadiq Hussain, Abbas Khosravi, U. Rajendra Acharya, Vladimir Makarenkov, Saeid Nahavandi

Machine Learning Faculty Publications

The COVID-19 (Coronavirus disease 2019) pandemic has become a major global threat to human health and well-being Thus, the development of computer-aided detection (CAD) systems that are capable to accurately distinguish COVID-19 from other diseases using chest computed tomography (CT) and X-ray data is of immediate priority Such automatic systems are usually based on traditional machine learning or deep learning methods Differently from most of existing studies, which used either CT scan or X-ray images in COVID-19-case classification, we present a simple but efficient deep learning feature fusion model, called UncertaintyFuseNet, which is able to classify accurately large datasets of …

Towards Improving Calibration In Object Detection Under Domain Shift, Muhammad Akhtar Munir, Muhammad Haris Khan, M. Saquib Sarfraz, Mohsen Ali

Computer Vision Faculty Publications

With deep neural network based solution more readily being incorporated in real-world applications, it has been pressing requirement that predictions by such models, especially in safety-critical environments, be highly accurate and well-calibrated. Although some techniques addressing DNN calibration have been proposed, they are only limited to visual classification applications and in-domain predictions. Unfortunately, very little to no attention is paid towards addressing calibration of DNN-based visual object detectors, that occupy similar space and importance in many decision making systems as their visual classification counterparts. In this work, we study the calibration of DNN-based object detection models, particularly under domain shift. …

Artificial Intelligence For Natural Disaster Management, Guansong Pang

Research Collection School Of Computing and Information Systems

Artificial intelligence (AI) can leverage massive amount of diverse types of data, such as geospatial data, social media data, and wireless network sensor data, to enhance our understanding of natural disasters, their forecasting and detection, and humanitarian assistance in natural disaster management (NDM). Due to this potential, different communities have been dedicating enormous efforts to the development and/or adoption of AI technologies for NDM. This article provides an overview of these efforts and discusses major challenges and opportunities in this topic.

Improving Data-Driven Infrastructure Degradation Forecast Skill With Stepwise Asset Condition Prediction Models, Kurt R. Lamm, Justin D. Delorit, Michael N. Grussing, Steven J. Schuldt

Faculty Publications

Organizations with large facility and infrastructure portfolios have used asset management databases for over ten years to collect and standardize asset condition data. Decision makers use these data to predict asset degradation and expected service life, enabling prioritized maintenance, repair, and renovation actions that reduce asset life-cycle costs and achieve organizational objectives. However, these asset condition forecasts are calculated using standardized, self-correcting distribution models that rely on poorly-fit, continuous functions. This research presents four stepwise asset condition forecast models that utilize historical asset inspection data to improve prediction accuracy: (1) Slope, (2) Weighted Slope, (3) Condition-Intelligent Weighted Slope, and (4) …

A Multi-Dimensional Matrix Pencil-Based Channel Prediction Method For Massive Mimo With Mobility, Weidong Li, Haifan Yin, Ziao Qin, Yandi Cao, Mérouane Debbah

Machine Learning Faculty Publications

This paper addresses the mobility problem in massive multiple-input multiple-output systems, which leads to significant performance losses in the practical deployment of the fifth generation mobile communication networks. We propose a novel channel prediction method based on multi-dimensional matrix pencil (MDMP), which estimates the path parameters by exploiting the angular-frequency-domain and angular-timedomain structures of the wideband channel. The MDMP method also entails a novel path pairing scheme to pair the delay and Doppler, based on the super-resolution property of the angle estimation. Our method is able to deal with the realistic constraint of time-varying path delays introduced by user movements, …

Forecasting Country Conflict Using Statistical Learning Methods, Sarah Neumann, Darryl K. Ahner, Raymond R. Hill

Faculty Publications

Purpose — This paper aims to examine whether changing the clustering of countries within a United States Combatant Command (COCOM) area of responsibility promotes improved forecasting of conflict. Design/methodology/approach — In this paper statistical learning methods are used to create new country clusters that are then used in a comparative analysis of model-based conflict prediction. Findings — In this study a reorganization of the countries assigned to specific areas of responsibility are shown to provide improvements in the ability of models to predict conflict. Research limitations/implications — The study is based on actual historical data and is purely data driven. …

Self-Supervised Video Object Segmentation Via Cutout Prediction And Tagging, Jyoti Kini, Fahad Shahbaz Khan, Salman Khan, Mubarak Shah

Computer Vision Faculty Publications

We propose a novel self-supervised Video Object Segmentation (VOS) approach that strives to achieve better object-background discriminability for accurate object segmentation. Distinct from previous self-supervised VOS methods, our approach is based on a discriminative learning loss formulation that takes into account both object and background information to ensure object-background discriminability, rather than using only object appearance. The discriminative learning loss comprises cutout-based reconstruction (cutout region represents part of a frame, whose pixels are replaced with some constant values) and tag prediction loss terms. The cutout-based reconstruction term utilizes a simple cutout scheme to learn the pixel-wise correspondence between the current …

Cost: Contrastive Learning Of Disentangled Seasonal-Trend Representations For Time Series Forecasting, Gerald Woo, Chenghao Liu, Doyen Sahoo, Akshat Kumar, Steven Hoi

Research Collection School Of Computing and Information Systems

Deep learning has been actively studied for time series forecasting, and the mainstream paradigm is based on the end-to-end training of neural network architectures, ranging from classical LSTM/RNNs to more recent TCNs and Transformers. Motivated by the recent success of representation learning in computer vision and natural language processing, we argue that a more promising paradigm for time series forecasting, is to first learn disentangled feature representations, followed by a simple regression fine-tuning step – we justify such a paradigm from a causal perspective. Following this principle, we propose a new time series representation learning framework for long sequence time …

Automation And Coupling Of Models For Coastal Flood Forecasting In South Texas, Cesar Davila Hernandez, Sara E. Davila, Martin Flores, Jungseok Ho, Dong-Chul Kim

Computer Science Faculty Publications and Presentations

Forecasting natural disasters such as inundations can be of great help for emergency bodies and first responders. In coastal communities, this risk is often associated with storm surge. To produce flood forecasts for coastal communities, a system must incorporate models capable of simulating such events based on forecasted weather conditions. In this work, a system for forecasting inundations based predominantly on storm surge is explored. An automation and a coupling strategy were implemented to produce forecasted flood maps automatically. The system leverages an ocean circulation model and a channel water flow model to estimate flood events in South Texas specially …

Seasonally And Diurnally Varying Cold Front Effects Along The Minnesotan North Shore Of Lake Superior, Matthew S. Van Den Broeke

Department of Earth and Atmospheric Sciences: Faculty Publications

Cold fronts are typically associated with cooling, drying and a strengthening wind that shifts to have a northerly component. Cold front effects at a particular point, however, are dependent upon pre-existing air mass characteristics. Here, we examine 634 passages of synoptic-scale cold fronts in northeastern Minnesota from 2010 to 2018. While these fronts are associated with the expected effects in some areas, they are often associated with warming and enhanced drying in the region directly influenced by an air mass from Lake Superior (coastal sites). Coastal sites experience warming during more than half of cold frontal passages, in contrast to …

Modeling Digital Camera Monitoring Count Data With Intermittent Zeros For Short-Term Prediction, Eben Afrifa-Yamoah, Ute Mueller

Research outputs 2022 to 2026

Digital camera monitoring has revolutionised survey designs in many fields, as an important source of information. The extended sampling coverage offered by this monitoring scheme makes it preferable compared to other traditional methods of survey. However, data obtained from digital camera monitoring are often highly variable, and characterized by sparse periods of zero counts, interspersed with missing observations due to outages. In practice, missing data of relatively shorter duration are mostly observed and are often imputed using interpolation techniques, ignoring long-term trends leading to inherent estimation biases. In this study, we investigated time series forecasting methods that adequately handle intermittency …

Predicting Pair Success In A Pair Programming Eye Tracking Experiment Using Cross-Recurrence Quantification Analysis, Maureen M. Villamor, Maria Mercedes T. Rodrigo

Department of Information Systems & Computer Science Faculty Publications

Pair programming is a model of collaborative learning. It has become a well-known pedagogical practice in teaching introductory programming courses because of its potential benefits to students. This study aims to investigate pair patterns in the context of pair program tracing and debugging to determine what characterizes collaboration and how these patterns relate to success, where success is measured in terms of performance task scores. This research used eye-tracking methodologies and techniques such as cross-recurrence quantification analysis. The potential indicators for pair success were used to create a model for predicting pair success. Findings suggest that it is possible to …

Atlantic Ocean Variability And European Alps Winter Precipitation, Giuseppe Formetta, Jonghun Kam, Sahar Sadeghi, Glenn Tootle, Thomas Piechota

Biology, Chemistry, and Environmental Sciences Faculty Articles and Research

Winter precipitation (snowpack) in the European Alps provides a critical source of freshwater to major river basins such as the Danube, Rhine, and Po. Previous research identified Atlantic Ocean variability and hydrologic responses in the European Alps. The research presented here evaluates Atlantic Sea Surface Temperatures (SSTs) and European Alps winter precipitation variability using Singular Value Decomposition. Regions in the north and mid-Atlantic from the SSTs were identified as being tele-connected with winter precipitation in the European Alps. Indices were generated for these Atlantic SST regions to use in prediction of precipitation. Regression and non-parametric models were developed using the …

Development Of Predictive Models For Water Budget Simulations Of Closed-Basin Lakes: Case Studies Of Lakes Azuei And Enriquillo On The Island Of Hispaniola, Mahrokh Moknatian, Michael Piasecki

Publications and Research

The historical water level fluctuations of the two neighboring Caribbean lakes of Azuei (LA) and Enriquillo (LE) on Hispaniola have shown random periods of synchronous and asynchronous behaviors, with both lakes exhibiting independent dynamics despite being exposed to the same climatic forces and being directly next to each other. This paper examines their systems' main drivers and constraints, which are used to develop numerical models for these two lakes. The water balance approach was employed to conceptually model the lakes on an interannual scale and examine the assumptions of surface and subsurface processes. These assumptions were made based on field …

Quo Vadis Lakes Azuei And Enriquillo: A Future Outlook For Two Of The Caribbean Basin's Largest Lakes, Mahrokh Moknatian, Michael Piasecki

Publications and Research

Lakes Azuei (LA) and Enriquillo (LE) on Hispaniola Island started expanding in 2005 and continued to do so until 2016. After inundating large swaths of arable land, submerging a small community, and threatening to swallow a significant trade route between the Dominican Republic and Haiti; worries persisted at how far this seemingly unstoppable expansion would go. The paper outlines the approach to a look forward to answer this question vis-à-vis climate change scenarios developed by the Intergovernmental Panel on Climate Change (IPCC). It uses numerical representations of the two lakes, and it examines how the lakes might evolve, deploying three …

We’Re Here To Get You There: A Statistical Analysis Of Bridgewater State University’S Transit System, Abigail Adams

Honors Program Theses and Projects

Bridgewater State University first established its on-campus transportation service in January of 1984. While it began only running as an on-campus service for students throughout the day, the service grew to expand by offering an off-campus connection to the neighboring city of Brockton and absorbed the night service system from the campus safety team. As BSU Transit continues to grow, the organization is seeking ways to improve their overall service and better prepare their fleet and driver pool to accommodate this growth. The purpose of this research is to analyze trends among the data collected by BSU Transit and assist …

Time Series Forecasting Of Covid-19 Deaths In Massachusetts, Andrew Disher

Honors Program Theses and Projects

The aim of this study was to use data provided by the Department of Public Health in the state of Massachusetts on its online dashboard to produce a time series model to accurately forecast the number of new confirmed deaths that have resulted from the spread of CoViD-19. Multiple different time series models were created, which can be classified as either an Auto-Regressive Integrated Moving Average (ARIMA) model or a Regression Model with ARIMA Errors. Two ARIMA models were created to provide a baseline forecasting performance for comparison with the Regression Model with ARIMA Errors, which used the number of …

Forecasting Of The Covid-19 Epidemic: A Scientometric Analysis, Pandri Ferdias, Ansari Saleh Ahmar

Library Philosophy and Practice (e-journal)

This study presented a scientometric analysis of scientific publications with discussions of forecasting and COVID-19. The data of this study were obtained from the Scopus database using the keywords: ( TITLE-ABS-KEY (forecast) AND TITLE-ABS-KEY (covid)) and the data were taken on March 26, 2021. This study was a scientometric study. The data were subsequently analyzed using the VosViewer and Bibliometrix R Package. The results showed that “COVID-19” was the keyword most frequently used by researchers, followed by “forecasting” and “human”. Authors who discussed the topic of forecasting COVID-19 come from 83 different countries/regions, with the most articles sent by authors …

Improving Stock Trading Decisions Based On Pattern Recognition Using Machine Learning Technology, Yaohu Lin, Shancun Liu, Haijun Yang, Harris Wu, Bingbing Jiang

Information Technology & Decision Sciences Faculty Publications

PRML, a novel candlestick pattern recognition model using machine learning methods, is proposed to improve stock trading decisions. Four popular machine learning methods and 11 different features types are applied to all possible combinations of daily patterns to start the pattern recognition schedule. Different time windows from one to ten days are used to detect the prediction effect at different periods. An investment strategy is constructed according to the identified candlestick patterns and suitable time window. We deploy PRML for the forecast of all Chinese market stocks from Jan 1, 2000 until Oct 30, 2020. Among them, the data from …

Stock Trend Prediction Using Candlestick Charting And Ensemble Machine Learning Techniques With A Novelty Feature Engineering Scheme, Yaohu Lin, Shancun Liu, Haijun Yang, Harris Wu

Information Technology & Decision Sciences Faculty Publications

Stock market forecasting is a knotty challenging task due to the highly noisy, nonparametric, complex and chaotic nature of the stock price time series. With a simple eight-trigram feature engineering scheme of the inter-day candlestick patterns, we construct a novel ensemble machine learning framework for daily stock pattern prediction, combining traditional candlestick charting with the latest artificial intelligence methods. Several machine learning techniques, including deep learning methods, are applied to stock data to predict the direction of the closing price. This framework can give a suitable machine learning prediction method for each pattern based on the trained results. The investment …

Changes In The Philippine Coastal Environment, Karl H. Szekielda, Ma. Aileen Leah G. Guzman

Environmental Science Faculty Publications

Global warming is progressing at a faster speed than has been estimated earlier in climate forecasting, and the ocean responds rather quickly to global temperature increase. This study uses remotely sensed data that were accessed from the System for Multidisciplinary Research and Applications (NASA Giovanni) to study environmental change in the Philippines’ coast. Monthly averaged sea surface temperature series from around the Philippines indicate that the Philippines follow the global trend in ocean temperature increase and show the increase of about 0.50C within two decades. Despite the high variability in temperature, the linear regressions displayed for all seasons show an …

Cost Estimating Using A New Learning Curve Theory For Non-Constant Production Rates, Dakotah Hogan, John J. Elshaw, Clay M. Koschnick, Jonathan D. Ritschel, Adedeji B. Badiru, Shawn M. Valentine

Faculty Publications

Traditional learning curve theory assumes a constant learning rate regardless of the number of units produced. However, a collection of theoretical and empirical evidence indicates that learning rates decrease as more units are produced in some cases. These diminishing learning rates cause traditional learning curves to underestimate required resources, potentially resulting in cost overruns. A diminishing learning rate model, namely Boone’s learning curve, was recently developed to model this phenomenon. This research confirms that Boone’s learning curve systematically reduced error in modeling observed learning curves using production data from 169 Department of Defense end-items. However, high amounts of variability in …