Educational Data Visualization Project
Authors: John E. Moser, Amy Gray, Laura Vavrek, Billy Musgraves, John P. Hogan, Jewell Simon, Zhaoxia "Mary" Wang, Wayne Picou
Special Thanks to: Dr. James Madden and Ameziane Harhad
There exist a large collection of data on the Louisiana's Department of Education Website concerning test scores that is in a raw format and divided amongst various excel files and PDF documents. The project we have created is a way to take this large amount of data and allowing the user to perform analysis on it with relative ease and efficiency in order to gain perspective on any particular subset of schools.
In order to accomplish this we inventoried the data and then ran it through various text editing scripts to delete special characters, correct any unusual formatting, and to correct any mistakes and then converted everything into CSV files that effectively created our database that can be easily used by our program.
Next, we performed some mathematical analysis to give more meaning to the data. In particular we gave the schools and test scores a one number summary that is used to rank these schools. We thought this score would give better insight than the current School Performance Score (SPS) that is currently in place.
Finally, we created a user interface that takes the data from our database and displays it in a graphical way so that those with non-mathematical backgrounds can easily compare schools and other aggregations of the data. We catered our design around administrators so they can use it to make improvements to their schools as they see fit. We used the programming language R to create a simple to use selection tool that when selected displays the information dynamically.
Measuring the Suck-Sess of Proper Orthogonal Decomposition on Data Obtained From An Infant Suck Apparatus
Authors: Zachary Bradshaw, Matthew Cavell, Murad Chowdhury, Jeannie Grodner, Joe Wilkerson Project Manager: Richard Frnka
Pennington Research is using a device to test the sucking methods of infants to determine if there is any link between suck rate and childhood obesity. In the summer of 2016 project, an interface was created to analyze the data and determine features of the tested information. The main technique to detect sucks was to compare a prototype of a typical suck against potential sucks in the data. The original prototype is merely a parabola and was used by the authors of the original process used to analyze the experimental data.
The prototype creator allows the user to manually choose sucks that represent the characteristics of the data set. Then principal component analysis is used to turn these sucks into information that can be used against other experiments for suck detection.
As a continuation from the project done in Summer 2016, we made improvements on the previous interface for suck detection and prototype creation process. The improvements will now detect almost all of the sucks in a test with nearly zero false positive responses, as compared with the original methodology. As sucks will generally have a well-defined peak, a peak-to-peak method is also implemented that can detect sucks in a different fashion. Additional features have been added to allow the research specialist running the tests to now manually decipher sucks based on graphical intuition. Extra prototypes have been created to allow comparisons and account for wide variability of the different infants partaking in the experiments.