MASTERS PRESENTATIONS FALL 2017 Thursday, December 14, 2017 9:00 - - PDF document

MASTER’S PRESENTATIONS

FALL 2017

Thursday, December 14, 2017 9:00 am – 12:00 pm Room KC 2204

SCHOOL OF CIS WINTER 2017 MASTERS PRESENTATIONS Thursday, December 14, 2017

Schedule of Presentations KC 2204: 9:00 am - Three to Five Minute Lightening Rounds David Meyer – MS Thesis Proposal, Advisor: Dr. Gregory Schymik “Developing 5GL Concepts from Pair Programming Interactions” Jadhav Amarnath – MS Project, Advisor: Dr. D. Robert Adams “FOODSHARE” Vinvith Kumar Mudugonda – MS Project, Advisor: Dr. D. Robert Adams “Catering Orders Management System” Abhinandan Vidya – MS Project, Advisor: Dr. D. Robert Adams “Database Synchronization in a Server Client Model” Clemencia Reyes Flores – MS Project, Advisor: Dr. D. Robert Adams “An Exploration of FileMaker Platform for Customs Compliance and Reporting” Matthew Englehart – MS Project, Advisor: Dr. Greg Wolffe “HAPPy: Home Affordability Predictor in Python” Moeen Farasat – MS Project, Advisor: Dr. Jerry Scripps “Real Time Visualization and Analysis of Tweets” Ryan Norton – MS Project, Advisor: Dr. Jonathan Leidig “Dynamic Database Schemas and Multi-Paradigm Persistence Transformations” David Rynties – MS Project, Advisor: Dr. Jonathan Leidig “Open Data: A User-Owned Centralized Data Repository” Matías Gil-Echavarría – MS Project, Advisor: Dr. Jonathan Engelsma “Using A Smartphone to Monitor Varroa Destructor in Honey Bee Colonies” Kyle Prins – MS Project, Advisor: Dr. Vijay Bhuse “14 Days of Vacation: A Rogue Switch Detection Technique” Juan Cárcamo Zuluaga – MS Thesis Proposal, Advisor: Dr. Greg Wolffe “Search-and-Rescue: Using Machine Learning to Develop Intelligent Unmanned Aerial Vehicles” Sixty-minute poster presentations to immediately follow

Developing 5GL Concepts from Pair Programming Interactions

Master’s Thesis Proposal

Presented By: David Meyer Advisor: Dr. Gregory Schymik

Abstract:

In the fulfilling of the contracts generated in Test Driven Development, a developer could be said to act as a constraint solver, similar to those used by a 5th Generation Language.(5GL) We, therefore, hypothesize that Fifth Generation linguistic mechanics, such as facts, rules, and goals, will be emergent in communications for a pair of developers performing Test Driven Development, validating 5GL syntax as congruent in the ways that practitioners communicate. Along the way, nomenclatures and linguistic patterns may be

• bserved which could inform the design of future 5GL languages.

FOODSHARE

Masters Project

Presented By: Jadhav Amarnath Advisor: Dr. D. Robert Adams

Abstract: Minimizing the food wastage by sharing food with people who are in need without any expenses through an app that is created using progressive web app technology. The main purpose of the project is to analyze the untouched food (hotel/catering) in an area and making an application to make the food available to the public. Through this application the food can be taken for no cost to anyone who needs it. This is done through a recent technology called “progressive web app technology”, which is a combination of web presence and native app. This app also includes map search allow users to easily search posts on map with marker information on it. Also push notification is supported when user is subscribed to it which will notify the user whenever there is post created.

Catering Orders Management System

Masters Project

Presented By: Vinvith Kumar Mudugonda Advisor: Dr. D. Robert Adams

Abstract: GVSU Campus Catering still uses a paper-based notice board for orders. The main challenge with this format is that changes to the order arrive up until the orders is finalized. This project describes the design and implementation of a web application for managing catering orders. This web application allows users to monitor upcoming orders, and log previous orders. In addition, the application provides notifications to the different users of the system.

Database Synchronization in a Server Client Model

Masters Project

Presented By: Abhinandan Vidya Advisor: Dr. D. Robert Adams

Abstract: Small mobile oriented applications usually have a small database that interacts with a central server synchronizing the data when the mobile device is online. There are many NoSQL options available on the market but none are simple enough for use as an embedded database and interact with standalone Relational Database Management Systems like MySQL or Oracle. The goal of the project is to understand how synchronization is achieved in the case of distributed database systems through a simple implementation. An application was developed using a server-client model with MySQL database on the server side and SQLite on the client side. The control of consistency and availability was baked into the application on both the Android and the Java server applications. By limiting the cache on the client, a virtual storage based limit was set on how long the offline mode can be continued in the absence of network connectivity.

An Exploration of FileMaker Platform for Customs Compliance and Reporting

Masters Project

Presented By: Clemencia Reyes Flores Advisor: Dr. D. Robert Adams

Abstract: This project aims to build a customs application named “CustCreator”. The customs process must first be explained to describe the function that the “CustCreator” provides. When foreign merchandise reaches the United States territory, those goods must be cleared through the U.S. custom authorities in order for them to be released and removed from the U.S. Customs

• warehouse. Every arrival of merchandise within U.S Customs territory must be supported by a form of

evidence of the right to make an entry. The process of entry is addressed by the submission of a unique import/export customs package that includes the Bill of Lading (for ocean shipments) or Air waybill, invoices, and duty forms. Documentation must be precise. Any slight discrepancies or omissions may prevent the merchandise from being cleared which results in a nonpayment, or may even result in the seizure of the importer’s goods by U.S Customs or foreign customs officials. The steps involved in designing the “CustCreator” platform begins with analyzing and comparing current customs technologies followed by a brief examination of several RDBMS packages namely, MySQL, PostgreSQL, Microsoft SQL and SQL Server. In addition, an assessment methodology was used for this project: Acceptance Criteria, Cost, IT Implementation, Assumptions and Deliverables. Therefore, the main functionalities of the customs database include assisting import/export brokerage

• perations in transmitting data applications to US Customs, creating and generating dynamic reports,

and building application forms to be processed online to a customs entity.

HAPPy: Home Affordability Predictor in Python

Masters Projects

Presented By: Matthew Englehart Advisor: Dr. Greg Wolffe

Abstract: From a credit and income perspective, the current home lending decision-making process is driven primarily by assessing a prospective borrower’s ability to repay a loan. Although effective from a credit risk perspective, this approach falls short of helping borrowers understand the much more nuanced question of how much they can afford to spend on a house. The current approach does not consider individual borrower preferences including savings and retirement goals and lifestyle choices. Lenders have an opportunity to develop a more guided “affordability” focused home lending experience by leveraging data that is readily available – including historical loan application data and deposit account transaction history. This research project used the “Fannie Mae Single-Family Loan Performance Data” dataset to create a proof-of-concept home affordability prediction model. Four classifiers were implemented and assessed to determine their suitability as prediction models: a logistic regression classifier, a polynomial regression classifier, a deep neural network classifier and a random forest classifier. Several techniques were leveraged to process the Fannie Mae data and optimize model performance including synthetic minority oversampling, feature scaling / normalization, feature engineering, k-fold cross validation and grid search. Two primary approaches were explored: using loan default status as the predictor of affordability and using monthly delinquency status to compute a custom affordability score that could be used as a

• predictor. Using the custom affordability scores binned into 4 classes as a predictor of affordability, the

random forest classifier was able to achieve an accuracy of 96.36%, with the lowest-scoring class achieving a prediction accuracy of 92%.

Real Time Visualization and Analysis of Tweets

Masters Project

Presented By: Moeen Farasat Advisor: Dr. Jerry Scripps

Abstract: Develop an application or a tool that will enable a user to get a bigger picture of what is happening around GVSU using twitter as social media. The idea behind this approach is to analyze and study tweets that are being posted by followers of Grand Valley State University. The medium of sharing information has changed rapidly over the last few years. Social network website like Facebook and Twitter are being used in a way that shape politics, business, world culture, education, careers, innovation, and more. Considering this huge impact of social networking website, my goal in this project was to gather data from twitter in real time. For limiting the scope of this project, I decided to collect only tweets from the users that are following Grand Valley State

• University. This tool will help the end user understand a community around a seed node by finding

most popular people, hastags tweets and more.

Dynamic Database Schemas and Multi- Paradigm Persistence Transformations

Masters Project

Presented By: Ryan Norton Advisor: Dr. Jonathan Leidig

Abstract: Today, countless businesses use relational databases to store essential information. That data, however, does not always come in the same structure. XML files, for example, may have various schemas for a document type, validated by multiple XSD files. It may not always make sense to use a traditional relational database for this storage, as NoSQL solutions offer flexibility, speed, and powerful tools (for search, analysis, and visualization). These documents and schemas are known and used by numerous branches or offices in a company but need to be stored in a centrally-located

• database. The goal of this work is to solve the problem of transforming XML files of various schema

types in the same database, by dynamically altering the schema of the database to accommodate the new file structures. In addition to relational database storage, the XML files are also mapped to a graph database to accommodate additional business needs such as visualizing relationships among the data using more powerful methods than traditional data stores. This project also aims to minimize the effort spent by a software developer persisting data with different schema types as well as creating methods for storing newly added schemas to the existing data persistence workflow. It achieves this by automating the process, using several existing persistence frameworks such as Java Architecture for XML Binding (JAXB), Hibernate Object-Relational Mapping (ORM), and the Neo4J Object Graph Mapping Library (OGM). This work integrates these technologies into a cohesive, configurable, highly-extensible framework that provides a largely automated solution to mapping evolving data structures to different data persistence architectures.

Open Data: A User-Owned Centralized Data Repository

Masters Project

Presented By: David Rynties Advisor: Dr. Jonathan Leidig

Abstract: A centralized data repository allows for accurate, consumer-generated data to be accessed by third parties while eliminating the need for data redundancies and inaccurate information. Current data strategies force multiple companies to request, store, protect, and maintain accurate records on each user - causing data redundancy, inaccuracy, increased risks, and a lack of transparency. With human- centric data quickly becoming more important in decision-making, it is imperative that this information becomes democratized to act as a catalyst for innovation and research while increasing the transparency of its use. Open Data was created to serve as a centralized repository of user data, using three separate AWS instances to allow for scalability, visualization, and reconfiguration as data grows. Based on user-contributed personal information, Open Data provisions content for 3rd party entities to purchase and access using microtransactions (e.g., political polling and surveys) or fulfill value-added requests from users (e.g., mortgage preapproval).

Using A Smartphone to Monitor Varroa Destructor in Honey Bee Colonies

Masters Project

Presented By: Matías Gil-Echavarría Advisor: Dr. Jonathan Engelsma

Abstract: Varroa mite, also known as varroa destructor (1), is a parasitic mite that infests honey bee colonies and transmit viruses to the bees, and is the major cause of bee disease and bee population decline (2). There is an effective method for varroa mite monitoring that can help quantify the extent of an infection. Some beekeepers are implementing this method, but they are not performing the process correctly and getting erroneous results that exacerbates the problem. One of the steps of the process requires the person to agitate a jar vigorously up and down for at least one minute. For an effective count, it is critical that the duration and rigor of the shaking motion is correctly performed. A mobile application using device accelerometers was implemented to teach beekeepers to understand and perform the method correctly. Additional research was done to use computer vision to count the removed mites automatically. And for the title, actually this extra time I am taking is because if I do not get something extra done (what I am trying to do right now), we will slightly change the title. So far it is: "Using smartphone to monitor varroa destructor in honey bee colonies.

14 Days of Vacation: A Rogue Switch Detection Technique

Masters Project

Presented By: Kyle Prins Advisor: Dr. Vijay Bhuse

Abstract: Networks are growing both in size and complexity. As this complexity increases it has become harder to detect a compromised node. In this presentation, we demonstrate an application of a business solution for detecting malicious

• actors. Our solution is based on an enforcement of 14 days of vacation per year. This is done in

banking as a best practice to detect employees committing fraud or otherwise engaging in a malicious behavior. We regularly swap nodes in the network with known good units and then compare observed behavior

• f the suspect node, to the behavior of the known good node. By doing this, we can detect many forms
• f malicious behavior indicating a compromised node.

Search-and-Rescue: Using Machine Learning to Develop Intelligent Unmanned Aerial Vehicles

Master’s Thesis Proposal

Presented By: Juan Cárcamo Zuluaga Advisor: Dr. Greg Wolffe

Abstract: Unmanned Aerial Vehicles (UAVs) are becoming more prevalent every day. Advances in battery life, sophisticated sensors, and reduced cost have spurred the development of many different applications

• utside the traditional military domain. For example, Search and Rescue (SAR) operations can now

use UAVs equipped with high-resolution cameras, enhancing their ability to conduct missions such as mapping, victim search, task observation, and early supply delivery. Although numerous unmanned systems (ground, aquatic, and aerial) have appeared in the last decade, this occurred before the ascendance of modern machine learning techniques; hence they rely heavily on greedy algorithms and potential-based heuristics. This research project will investigate enhanced approaches to SAR that incorporate recently discovered Reinforcement Learning methods. The experimental framework will utilize open-source tools such as Microsoft’s state-of-the-art flight simulator (AirSim) and Google’s machine learning framework (TensorFlow). Two different machine learning methods will be implemented: a Deep Recurrent Q- Network (DRQN), and the Asynchronous Advantage Actor Critic (A3C). The approaches will be compared and evaluated based on machine learning metrics and performance in common SAR tasks.