Identifying Robust Classifiers of Sequential Process Data in Adults' Digital Literacy Tasks: A Machine Learning Approach

• In Event: AERA Graduate Student Research-in-Progress Session 5
  In Roundtable Session: Digital Literacy and Digital Play: Content Creation, Instructional Strategies, and Learning Patterns (Table 11)

Abstract

This study used the process data recorded in a computer-based large-scale program, the Programme for International Assessment of Adult Competencies (PIAAC), to discover the relationship between behavioral action sequences and the task performances. The purpose of the study is twofold: first, to find the most influential combinations of two or three robust classifiers from time-embedded action sequences to distinguish correct/incorrect groups by using the N-grams method, and second, to find whether there are any associations between these robust classifiers and the participants' general reading skills. The methodologies used include random forest classifiers to select robust features and to make predictions, and chi-square selection to distinguish participants into different performance groups.

Author

• Yiming Chen, University of Minnesota