Does Difficulty even Matter? Investigating Difficulty Adjustment and Practice Behavior in an Open-Ended Learning Task

This repository contains data, code, and plots used in the paper "Does Difficulty even Matter? Investigating Difficulty Adjustment and Practice Behavior in an Open-Ended Learning Task". There are two main Python scripts for generating the results:

• compare-conditions.py - Generates plots comparing the different measures between the conditions.
• clustering.py - Reads the practice behaviors of the students, then clusters the students based on that. The script outputs plots comparing different measures between the clusters, and performs Kruskal-Wallis tests. Finally, the script also mines the association rules, associating the clusters with characteristics of the practice behavior. The script requires a JSON setting file, which is at plots/click_type_num_time/settings.json. In this case, we run the script with python clustering.py click_type_num_time.

The results from both scripts (and also are there without you having to run the scripts) are in the directory plot. The subdirectory conditions contains the plots comparing between conditions. The subdirectory click_type_num_time contains plots comparing the clusters, and a log output file log.json. The log contains the numerical results, the statistical tests, along with the extracted association rules of the clusters.

We recommend using venv, with the pip requirements provided requirements.txt.