Session Summary
Share...

Direct link:

Applying Data Mining Methods to Detect Test Fraud

Thu, April 11, 8:45am to 12:45pm, Convention Center, Floor: First, 122B

Session Type: Training Session

Abstract

This session will provide audience with systematic training on applying various data mining models using software programs: R and/or Python to detect fraud in different test formats, such as computer based, computer adaptive or multistage settings. It covers the basics of these two software programs, theories of selected unsupervised and supervised learning methods, including K-Means, Gaussian Finite Mixture, Self-Organization Mapping, K-Nearest Neighbor, Random Forest, Supported Vector Machine, Neural Network with R/Python demonstrations. Further, the advantages and disadvantages of using each software program will be discussed.

This session consists of lectures, demonstrations, and hands-on activities of running various commonly used data mining methods. It is intended for intermediate and advanced graduate students, researchers, and practitioners who are interested in learning the basics and advanced topics related to data mining methods. It is expected the audience will have some basic knowledge of R and Python programming, but not required. Attendees will bring their own laptop and download the software programs free online. It is expected that attendees will master the basics of specify various data mining models and applying these models to detect aberrantly behaved test-takers; further, they can apply the skills to their own research and datasets.

Sub Unit

Session Organizers

Presenter