Activity

Participants generate a synthetic dataset, build the response matrix, inspect CLAIRE quantities, and work through the main activity of the integrated final block.

Browser Lab

This page includes executable Python cells directly in the site. When a notebook backend is connected, birt is available and the full CLAIRE workflow can run here; otherwise the page falls back to the pure-Python workshop helpers.

Shared Helpers You Can Use

The CLAIRE activity can use the same helpers in the browser notebook, in the local notebook, and in Colab. The point is to let you focus on the latent analysis rather than rewriting support code.

from utils.handson import (
    beta4_expected_response,
    build_claire_response_matrix,
    compute_claire_like_scores,
    estimate_case_statistics,
    make_toy_clustering_partitions,
)
from utils.transform import TransformPairwise

• make_toy_clustering_partitions gives you a compact starter set of partitions for a CLAIRE-style demo.
• build_claire_response_matrix and TransformPairwise both make the agreement-to-response-matrix step explicit.
• estimate_case_statistics and compute_claire_like_scores provide simple summaries that help inspect the response matrix before or after fitting.
• beta4_expected_response is available when you want to interpret representative items back through a bounded-response curve.

Inside the browser notebook editor, hovering one of these helper names now shows its docstring.

Notebook

• Main final-block activity notebook: 05_02_activities.ipynb
• Goal: inspect individual items and connect the practical workflow back to the CLAIRE theory

What Participants Should Do

• generate a synthetic dataset;
• produce the response matrix;
• inspect ability, difficulty, discrimination, and ICCs.