Date of Award
Fall 2018
Document Type
Dissertation
Degree Name
Doctor of Philosophy (PhD)
Department
Mathematics, Statistics and Computer Science
Program
Computational Sciences
First Advisor
Ahamed, Sheikh I
Second Advisor
Smith, Roger O.
Third Advisor
Guastello, Stephen
Abstract
The processes of collecting, cleaning, and presenting data are critical in ensuring the proper analysis of data at a later date. An opportunity exists to enhance the data collection and presentation process for those who are not data scientists – such as healthcare professionals and businesspeople interested in using data to help them make decisions. In this work, creating an observational data collection and presentation tool is investigated, with a focus on developing a tool prioritizing user-friendliness and context preservation of the data collected. This aim is achieved via the integration of three approaches to data collection and presentation.In the first approach, the collection of observational data is structured and carried out via a trichotomous, tailored, sub-branching scoring (TTSS) system. The system allows for deep levels of data collection while enabling data to be summarized quickly by a user via collapsing details. The system is evaluated against the stated requirements of usability and extensibility, proving the latter by providing examples of various evaluations created using the TTSS framework.Next, this approach is integrated with automated data collection via mobile device sensors, to facilitate the efficient completion of the assessment. Results are presented from a system used to combine the capture of complex data from the built environment and compare the results of the data collection, including how the system uses quantitative measures specifically. This approach is evaluated against other solutions for obtaining data about the accessibility of a built environment, and several assessments taken in the field are compared to illustrate the system’s flexibility. The extension of the system for automated data capture is also discussed.Finally, the use of accessibility information for data context preservation is integrated. This approach is evaluated via investigation of how accessible media entries improve the quality of search for an archival website. Human-generated accessibility information is compared to computer-generated accessibility information, as well as simple reliance on titles/metadata. This is followed by a discussion of how improved accessibility can benefit the understanding of gathered observational data’s context.