Master's Presentations Computer Information Systems Fall 2019 - PDF document

Master's Presentations Computer Information Systems Fall 2019 Thursday, December 12, 2019 8:00-10:00a Grand River Room (2250 KC)

School of Computing and Information Systems Master’s Presentations Computer Information Systems Thursday, December 12, 2019 Schedule of CIS Presentations: 8:00 am - Three to Five Minute Lightning Rounds Brian Mbeere – MS Project, Advisor: Dr. Zachary Kurmas “Application of GPS Sensors and QR Codes in Yard Management” Lucas Crandle – MS Project, Advisor: Dr. Jared Moore “Analysis of Beekeeper Hive Event Annotations” Nathan Hull – MS Project, Advisor: Dr. Jonathan Engelsma “Improving Data Quality via User Engagement” John DeGrandchamp – MS Project, Advisor: Dr. Byron DeVries “MAGC Inc., Website Redevelopment” Jason Hansen – MS Project, Advisor: Dr. Byron DeVries “Development of an iPhone Application to Increase Usability of the Village of Caledonia Website” Joshua Spicer-Sweet – MS Project, Advisor: Dr. Byron DeVries “Creating an eCommerce Web Application Using Vue.js and Web API” Anne. Vamsypriya – MS Project, Advisor: Dr. Yonglei Tao “Secure Backup Software System” Sudhir Gaire – MS Project, Advisor: Dr. Yonglei Tao “ORL (Otorhinolaryngology) Patient Databank Application” Geethanjali Sanikommu – MS Project, Advisor: Dr. Yonglei Tao “Aves Classifier – An Image Recognition iOS Application” Sean Driscoll – MS Project, Advisor: Dr. Jagadeesh Nandigam “Blackboard Automated Testing Framework” Sneha Joshi – MS Project, Advisor: Dr. Jagadeesh Nandigam “Ridge Voice: A Community Association iOS App” Michael Palazzolo – MS Project, Advisor: Dr. Vijay Bhuse “Open Source Approach to SIEM Using the Elastic Platform” Ahmad Sheikh-Khalil – MS Project, Advisor: Dr. Vijay Bhuse “Combating Rogue Switches” Paige Melick – DSA Internship, Advisor: Dr. Jerry Scripps “Data Analyst Intern Experience at Herman Miller” Recodeo Rekod – MS Thesis Proposal, Advisor: Dr. Jerry Scripps “A Deep Learning Model to Classify Bone Fracture Healing Stage with Ultrasound Propagation” Poster presentations to immediately follow for the remaining time.

Application of GPS Sensors & QR Codes in Yard Management CIS Master’s Presentation Presented By: Brian Mbeere Advisor: Dr. Zachary Kurmas Abstract: Steelcase Inc. is faced with a challenge where they cannot locate some of their trailers once they are dropped off in their plant’s yard. The problem mainly faces trailers that are not owned by Steelcase since some third-party carriers drop off loads in lots that are not designated for their trailers. The current system assigns and records a lot number to a trailer at the gate which causes confusion if the truck driver does not drop the trailer in the specified lot. The project involved developing two solutions to this challenge which include; a GPS-based yard management system and a QR-based yard management system. The GPS based yard management system has a Nodejs backend which makes API calls to get locational data from a portal that the GPS sensors transmit data to. The backend then provides this data as a web service to a reacts frontend. The reacts frontend makes use of the google maps JavaScript API to display the data on a map and satellite image. The reacts frontend has a truck details page that enables creation of truck details which can be associated with a GPS sensor that is attached to a trailer for tracking purposes. When the sensor is displayed on the map, it shows the location of the trucks it is attached to. The QR-code based yard management system can be used by the staff tasked with shifting trailers from one lot to another. For the solution to be effective, Geo tagged QR codes would have to be erected next to lots in the yard and QR codes with truck details stuck on trucks upon entering the plant’s yard. The app would be used by the staff who are assigned the task of moving trailers from one lot to another. They would scan the Geo-tagged QR code and the trucks QR code every time they parked the trailers thus sharing its location. After a feasibility analysis of the two solutions, the GPS based system has higher hardware costs since each trailer that is being tracked must have a sensor attached to it. Its development cost is lower than the QR code since the company would have to outsource the mobile app development and maintenance which is expensive. Some of the challenges the GPS solution would face include; loss of sensors due to battery drain, damage of sensors could lead to cost increase while replacing them, sensors need to be in line of sight with the sky. The QR code solution would also experience challenges like, it is prone to human error where users can forget to scan the QR code and there is high initial cost of erecting QR codes on every parking space.

Analysis of Beekeeper Hive Event Annotations CIS Master’s Presentation Presented By: Lucas Crandle Advisor: Dr. Jared Moore Abstract: The Bee Informed Partnership (BIP) is a non-profit organization helping beekeepers improve the health and wellness of their bee colonies. One initiative supported here at GVSU is to provide hive management guidance through data collection and analysis. Currently, the GVSU team is developing web applications to support hive weight collection through sensors placed under hives. The team has developed a classification model to identify extraordinary change in hive weight, ultimately prompting beekeepers to annotate the reason for the weight change event. Often, these weight change events correspond to beekeeper actions such as hive maintenance, monitoring, and honey harvesting. As a complement to the deployed model, in this project, we investigate the beekeeper weight change annotations currently in the dataset to (1) assess the quality of the data for use in future machine learning models and (2) to provide recommendations for improving the web portal presented to beekeepers. With regards to annotation data quality, we have identified a few trends that we must address to improve data gathering efforts. Notably, the annotations vary significantly in length and content from hive to hive. Each beekeeper has their own writing style and descriptions, even for common maintenance activities such as feeding. The primary contribution of my work was thus to develop a process to standardize annotation content to a certain specific range of topics. This allows us to group similar annotations together for developing advanced models in the future. The process first employs a key-word search derived from common beekeeper actions, then applying an N-Gram technique to generate a list of common actions that beekeepers are carrying out on their hives. Through these techniques, I was able to extract a common list of actions that beekeepers are performing on their hives. This pipeline can now be integrated into the BIP web portal to facilitate data standardization enabling further machine learning approaches in the future.

Improving Data Quality via User Engagement CIS Master’s Presentation Presented By: Nathan Hull Advisor: Dr. Jonathan Engelsma Abstract: The Bee Informed Partnership currently collects data from over 1000 hive monitors, which send data such as weight and temperature from hives all around the country in varying intervals. Other developers at Grand Valley have designed ML algorithms to identify perturbations in data, or what we call predictions. The next step is getting users to label that data with information about the event, a process we call annotation. These events vary from honey harvesting to feedings, and event swarm events. In the future, this labeled data will allow the development of ML-driven annotation, and one day, predictive warnings. As the volume and velocity of data continue to grow, so too does the value in leveraging it. A key factor of that is user engagement — the user finding value in your product, providing invaluable metrics along the way. Here, we sought to increase that engagement by designing a new dashboard, with an emphasis on focus, direction, and ease-of-use.

MAGC Inc., Website Redevelopment CIS Master’s Presentation Presented By: John DeGrandchamp Advisor: Dr. Byron DeVries Abstract: The Michigan Association of Genetic Counselors’ website was failing to meet the organization’s needs on multiple levels. Primarily, the prior site did not provide enough flexibility in terms of managing and adding content on the various pages. Additionally, the site was not entirely mobile friendly. In 2019, with a large portion of traffic coming through mobile devices, not being mobile friendly hampered usability. This project is focused on creating a new website for the Michigan Association of Genetic Counselors through the usage of a modern technology stack. This includes using Angular for the user interface implementation, Firebase to push typical server functions to the cloud, and Flamelink to provide a headless CMS to deliver the high levels of flexibility the organization desired. All key functions of the new site have been completed and planned go-live for conversion of www.magcinc.org from the old site to the new site is on or before December 31, 2019.