Python
Plotly/Dash
Overview
Open learner models make an educational system’s estimate of a learner visible to that learner. This project is a test bed for implementing those models, simulating interactions, and examining how changes in evidence affect estimated mastery over time.
The dashboard is designed for live experimentation rather than static reporting. Model parameters and learner behavior can be varied while the resulting knowledge state is visualized, making assumptions that are often hidden inside adaptive systems easier to inspect.
Current Model
The first implementation uses Bayesian Knowledge Tracing (BKT). BKT treats mastery as a hidden state and updates its probability after each correct or incorrect response. Parameters describe prior knowledge, learning, guessing, and slipping.
The simulation separates the mastery-model logic from the interface so additional models can be compared against the same interaction scenarios in the future.
Dashboard
The current application uses Plotly and Dash for its interactive Python interface and visualizations. It supports rapid local experimentation without requiring a separate JavaScript frontend or API service.
Potential uses include:
- Comparing learner-model behavior under controlled response sequences
- Demonstrating how noisy evidence changes mastery estimates
- Testing what information an open learner model should expose
- Prototyping dashboard explanations before integration into a learning platform