BEGIN:VCALENDAR
VERSION:2.0
PRODID:icalendar-ruby
CALSCALE:GREGORIAN
X-WR-CALNAME:M.S. Project Defense: Egill Gunnarsson
X-WR-TIMEZONE:Eastern Time (US & Canada)
BEGIN:VEVENT
DTSTAMP:20260908T190219Z
UID:tag:localist.com\,2008:EventInstance_37554345689687
DTSTART:20210819T140000Z
DTEND:20210819T150000Z
DESCRIPTION:Interactive Supervised Machine Learning Model Evaluation Using 
 D3\n\nEgill Gunnarsson\nAdvisor: Dr. Rafal Angryk\n\nThe evaluation of a s
 upervised machine-learning model is one of the most important phases in it
 s life cycle. Although there are numerous evaluation metrics\, each of whi
 ch provides a different insight into a model's performance\, it can someti
 mes be challenging to find the appropriate ones that fit the problem in ha
 nd. Including the imbalance ratio as an extra variable makes the evaluatio
 n process even more difficult. Therefore\, I implemented a web application
  to intuitively evaluate models' performance based on their confusion matr
 ices and given imbalance ratios. This project is an online\, interactive a
 pplication of the contingency space concept recently proposed by Ahmadzade
 h et al. (2021). Inspired by this concept\, my web application allows the 
 user to visually evaluate their pre-trained supervised models. A side-by-s
 ide graphical representation of multiple metrics is provided for a compari
 son between metrics scores. Confusion matrices are evaluated on such metri
 cs as accuracy\, precision\, F1 score\, and recall. Additionally\, the use
 r can load their own customized metrics as well. The visualization is base
 d on contour plots that correlate to each metrics score in relation to tru
 e positive and true negative rates and imbalance ratios.\n\nThis applicati
 on uses technologies such as D3.js\, Python\, JavaScript\, HTML\, CSS\, Fl
 ask\, and JSON. Each metric’s score is generated in the backend using Py
 thon\, based on an imbalance ratio. Information is sent to and from the ba
 ckend via Flask and JSON objects. JavaScript then uses the D3 library to c
 onvert the metric scores into a contour plot. The D3 library has many inte
 ractive capabilities\, allowing the user to modify the evaluation to fit e
 very requirement.\n\nCommittee\nDr. Rafal Angryk (chair)\nDr. Azim Ahmadza
 deh
LOCATION:
SUMMARY:M.S. Project Defense: Egill Gunnarsson
URL;VALUE=URI:https://calendar.gsu.edu/event/ms_project_defense_egill_gunna
 rsson
CATEGORIES:Science & Tech
CATEGORIES:Graduate Thesis Presentation
END:VEVENT
END:VCALENDAR