Adaptive testing moodle experiment

This repository contains minimal logic for adaptive testing using:

• IRT (Rasch Model)
• BKT (Bayesian Knowledge Tracing via pyBKT)

Structure

• models/bkt_model.py: Adaptive BKT model with skill tracking.
• models/irt_model.py: Simple IRT model with online theta updating.
• data/question_metadata.csv: Question skill and difficulty.
• data/response_logs.csv: User answer logs.
• requirements.txt: Install dependencies with pip install -r requirements.txt.

Example Usage

from models.bkt_model import BKTModel
from models.irt_model import IRTModel
import pandas as pd

metadata = pd.read_csv("data/question_metadata.csv")

# Initialize and train BKT
bkt = BKTModel(metadata)
bkt.add_response("user1", "q1", 1)
bkt.add_response("user1", "q2", 0)
bkt.train()
print("Weakest skill:", bkt.get_weakest_skill())

# Initialize and update IRT
irt = IRTModel(metadata)
irt.update("q1", 1)
irt.update("q2", 0)
print("Next IRT question:", irt.get_most_informative(["q1", "q2"]))