Echelon; meaningful feature extraction and clustering on SQL queries

Weston, Matthew Charles

Permalink

https://hdl.handle.net/2142/116293 Copy

Description

Title

Echelon; meaningful feature extraction and clustering on SQL queries

Author(s)

Weston, Matthew Charles

Issue Date

2022-07-22

Director of Research (if dissertation) or Advisor (if thesis)

Alawini, Abdu

Department of Study

Computer Science

Discipline

Computer Science

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

M.S.

Degree Level

Thesis

Keyword(s)

• SQL
• Clustering

Abstract

A core part of Computer Science education, and Database Systems education in particular, is the use of machine problems to both develop and assess students’ abilities. In order to conserve resources, these assignments are often automatically graded via auto-grading systems that verify that they produce the correct outputs. Unfortunately, these systems lack essential insights into the approaches students use to solve the assignments being graded, allowing subtle flaws in student intuition to go unseen. Furthermore, manual analysis of students’ code submissions at scale ranges from costly to impossible, depending on course size and assignment frequency, making these drawbacks difficult to avoid. In this thesis paper, we rigorously define a series of metrics for evaluating a system that captures nuances in students’ approach, and then make use of these metrics to develop a system that is capable of serving as a significant force multiplier for Computer Science faculty. This system, Echelon, functions by extracting features that instructors deem significant from students’ SQL queries and using them to generate clusters that capture the key approaches taken, and then projecting these clusters to an interactive dashboard that can be used to help teaching staff quickly identify the major trends in students’ approaches to a problem. We conclude with a full analysis of Echelon on real data.

Graduation Semester

2022-08

Type of Resource

Thesis

Copyright and License Information

Copyright 2022 Matthew Weston

Owning Collections