Analysis Recipes

Sequence Recipe

from design_research_analysis import fit_markov_chain_from_table

result = fit_markov_chain_from_table(rows, order=1, smoothing=1.0)
print(result.transition_matrix)

Model Comparison Recipe

from design_research_analysis import fit_markov_chain_from_table

baseline = fit_markov_chain_from_table(rows_a, order=1, smoothing=1.0)
variant = fit_markov_chain_from_table(rows_b, order=1, smoothing=1.0)

difference = baseline - variant
effect = baseline / variant
print(difference.to_dict())
print(effect.effect_size)

Language Recipe (Custom Embedder)

from design_research_analysis import compute_language_convergence

lookup = {
    "alpha": [2.0, 0.0],
    "beta": [1.0, 0.0],
    "gamma": [0.0, 0.0],
}

convergence = compute_language_convergence(
    rows,
    window_size=1,
    embedder=lambda texts: [lookup[text] for text in texts],
)
print(convergence.direction_by_group)

Embedding Maps Recipe

import numpy as np
from design_research_analysis import (
    build_embedding_map,
    cluster_embedding_map,
    compute_design_space_coverage,
    compute_divergence_convergence,
    compute_idea_space_trajectory,
)

vectors = np.asarray(
    [
        [1.0, 0.0, 0.1],
        [0.9, 0.2, 0.1],
        [0.0, 1.0, 0.2],
        [0.1, 0.9, 0.3],
    ]
)
embedding_map = build_embedding_map(vectors, method="pca", n_components=2)
clusters = cluster_embedding_map(embedding_map, n_clusters=2)
coverage = compute_design_space_coverage(embedding_map)
trajectory = compute_idea_space_trajectory(
    embedding_map,
    timestamps=[0, 1, 2, 3],
    groups=["session-a"] * 4,
)
dynamics = compute_divergence_convergence(trajectory, window=2)
print(clusters["labels"])
print(coverage["pairwise_spread"]["mean"])
print(dynamics["groups"]["session-a"]["dominant_direction"])

Statistics Recipe

from design_research_analysis import fit_regression

X = [[0.0], [1.0], [2.0], [3.0]]
y = [1.0, 3.0, 5.0, 7.0]
reg = fit_regression(X, y, feature_names=["x"])
print(reg.coefficients, reg.intercept, reg.r2)

Experiment Condition Comparison Recipe

from design_research_analysis import (
    build_condition_metric_table,
    compare_condition_pairs,
)

condition_metric_rows = build_condition_metric_table(
    runs_rows,
    metric="market_share_proxy",
    condition_column="selection_strategy",
    conditions=conditions_rows,
    evaluations=evaluations_rows,
)

report = compare_condition_pairs(
    condition_metric_rows,
    condition_pairs=[
        ("profit_focus_prompt", "neutral_prompt"),
        ("neutral_prompt", "random_selection"),
    ],
    alternative="greater",
    seed=17,
)
print(report.render_brief())
print(report.to_significance_rows())

Dataset And Provenance Recipe

import pandas as pd
from design_research_analysis import (
    capture_run_context,
    generate_codebook,
    profile_dataframe,
    validate_dataframe,
    write_run_manifest,
)

frame = pd.DataFrame({"participant_id": [1, 2, 3], "condition": ["A", "A", "B"]})
profile = profile_dataframe(frame)
validation = validate_dataframe(frame, {"participant_id": {"unique": True, "nullable": False}})
codebook = generate_codebook(frame)
context = capture_run_context(seed=7)
write_run_manifest(context, "artifacts/run_manifest.json")
print(profile["n_rows"], validation["ok"], list(codebook["column"]))