Big Data and Analytics: Pathways to Maturity

Permanent URI for this collectionhttps://hdl.handle.net/10125/107419

Browse

Now showing 1 - 9 of 9

Interoperability for Autonomy
(2024-01-03) Combs, Kara; Bihl, Trevor; Pennington, James
Autonomous systems aim to augment human capabilities with machine-based decision-making in the absence of a user. Ideally, autonomy hardware and software would be modular, having the ability to swap components in and out as needed based on necessary capabilities. However, many legacy systems in use utilize proprietary software with specific standards and components, reducing the system’s ability to be interoperable. Currently, the literature’s definition of interoperability is vague and often mistaken for other similar terms. We distinguish the uniqueness of interoperability and codify it through a taxonomy. Next, we extend this framework to understand autonomy and its hardware/software components through a proposed unified autonomy stack. We then evaluate the similarity between four autonomy architectures based on 29 stack components that are later presented in the “interchangeability matrix.” Thus, we demonstrate the necessity to unify autonomy hardware/software under the proposed taxonomy in the development of future autonomous systems.
Analytics Morphology and Transformation
(2024-01-03) Kaisler, Stephen; Money, William; Espinosa, J. Alberto; Armour, Frank
Earlier research in Big Data and Analytics led to the development of an analytic class taxonomy which organized analytics by application area. Further research motivated a need for better descriptions and a realization that analytics have many variations based on implementations for specific platforms. The authors realized that transformations of an analytic from one form to another could yield improvements in performance, space utilization, and other attributes. This paper examines analytic forms and the transformations among them with the goal of documenting them in a comprehensive manner. It proposes the development of an Analytics Catalog as a mechanism for documenting analytics classes, their membership, and the transformations among forms.
Classification of Experience for Proactive In-Car Function Recommendations Based on Customer Usage Data
(2024-01-03) Micus, Christian; Homola, Daniel; Böhm, Markus; Krcmar, Helmut
Automotive companies can use data from connected vehicles to enhance customer experience. Driver assistance functions have a low usage rate, and appropriate proactive function recommendations can improve both usage rate and customer experience. Qualitative studies often drive the development, and functions are recommended using a rule-based system. We provide a patented machine learning-based classification concept to make intelligent function recommendations based on customer usage. Therefore, we classify customer experience based on the driving context. We defined how to create an experience label for a function activation context and evaluated the approach using 716,000 function activations collected from the customer fleet data by an automotive manufacturer. To improve the quality of the binary classification model, we defined geospatial key performance indicators that provide quantifiable measures for the performance of a function on a road section. Our results reveal that the novel classification concept is a viable solution for car function recommendations.
Data Analytics Capability Maturity Models for Small and Medium Enterprises – A Systematic Literature Review
(2024-01-03) Marohn, Robert; Li, Yan
Organizations are recognizing the importance of investing in data analytics. For small and medium-sized enterprises (SMEs), developing effective data analytics capabilities can be an overwhelming task due to their limited resources. A Data Analytics Capability Maturity Model (DACMM) can be an essential tool for SMEs to assess their data analytics capabilities, pinpoint improvement areas, and create a roadmap for enhancing their data analytics maturity. Thus, a systematic literature review (SLR) is used to assess if any existing models may address the needs of SMEs in developing their data analytics capabilities. The SLR reviewed 18 models based on their component characteristics, assessment approaches, prescriptive focus, and theory and methodology used in model development. The result shows gaps in existing models, including limited dimensions for data management, the absence of a prescriptive model, and the need for theory-based, evidence-informed model development that fits the specific needs of SMEs. To this end, this research calls for the development of a new DACCM for SMEs using action design research.
Impact of Sustainability Disclosures on Financial Performance: A Natural Language Processing Perspective
(2024-01-03) Lui, Gladie; Chia, Jeremy; Shum, Connie
Sustainability reporting has emerged as an increasingly important topic since the trend linking financial and non-financial performance in the capital market attracted great attention in recent years. This study employs Natural Language Processing (NLP) to investigate three characteristics of sustainability reports including major component topics, readability, and sentiments of words used. Findings from the NLP analysis are regressed against corporate financial performance measures to examine the relation between sustainability report characteristics and firm performance. Results show that specific thematic topics of sustainability reports relate to different financial performance indicators. For instance, community-related disclosures have a positive relation with Return on Assets (ROA) and with Market Value (MV), methodology-related topics have a positive relation with MV, resource-related topics have a negative relation with MV, and governance and climate-related topics both have a positive relation with the Zmijewski score (ZJS), suggesting that these factors contribute to a firm’s financial viability.
Exploring the Intellectual Composition of Academic Research Conferences: Computational Text Analysis of the HICSS Paper Archive from 2017-2022
(2024-01-03) Cogburn, Derrick; Ochieng, Theodore; Buehlman Barbeau , Sierra; Wong, Haiman
Academic research conferences play a critical role in national and international scientific production. For example, the Hawaii International Conference on System Sciences is one of the longest running academic conferences in the world. HICSS consistently produces a wide range of high-quality, peer-reviewed research papers, distributed amongst 10 core tracks, and multiple minitracks. This paper provides a computational method for assessing academic research conferences by exploring the intellectual composition of the HICSS conference asking: what themes are most prevalent across the conference? Are topics identifiable? Can we predict the track of a paper from its abstract? To answer these questions, we analyze the HICSS papers from 2017-2022 (n=5,024). Applying inductive and deductive text-mining techniques, including: “Bag of Words” frequencies, NLP, unsupervised and supervised machine learning, we find several consistent themes and topics over the past five years, as well as meaningful divergence. Finally, the abstract of a paper predicts its track.
A Graph-Theoretic Approach for Examining Team Communities in Sports Transfer Markets
(2024-01-03) Lillo Portero, Eleuterio; Ali, Hesham
The sport transfer market is a designated timeframe where teams engage in player acquisitions and sales, aiming to enhance their rosters. The transfer process has been growing significantly in the last few years. Therefore, the goal of this study is to investigate the intricate dynamics of both local and global transfer markets, treating them as complex networks, to comprehend their influence on club performances. We employ graph theoretic concepts to model the transfer process and analyze its outcomes. To illustrate our approach and assess its results, we have constructed a complex network based on European soccer leagues, focusing on the specific case study of the English Premier League as one of the most recognized professional leagues in the world.
The Effectiveness of Fitness Wearable Technologies: Applying Big-Data Techniques to Analyze User Experience and Perception
(2024-01-03) Sharma, Arpit; Joshi, Deepti; Money, William; Mcwilliams, Denise
Fitness Wearable Technologies (FWT) have been analyzed as input devices for various user outcomes. However, user motives behind device usage have not been assessed from a longitudinal or stability perspective, nor have they been compared across different devices. FWT usage is frequently assumed and analyzed from the usage (act of wearing) of the device. This paper examines user motivations by analyzing user perceptions across three themes: Accomplishment (fitness goals), Health (chronic), and Social (connections with peers), espoused by wearers. The analysis is based upon Twitter theme data drawn in the samples from the spring quarter of ‘Pre, During, and Post’ “Covid” years (2019–2022). The findings reveal features of FWTs that motivate users and that users continue to embrace the use of FWTs for medical advantages and social and personal development. Additionally, sentiment analysis of the tweets across devices over the study period reveals an increase in the positive sentiment.
Introduction to the Minitrack on Big Data and Analytics: Pathways to Maturity
(2024-01-03) Kaisler, Stephen; Armour, Frank; Espinosa, J. Alberto; Cogburn, Derrick

Browse

Recent Submissions