Spaces:
Sleeping
Sleeping
File size: 5,833 Bytes
3a41bf2 471a2fe 5fbbdae 3a41bf2 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 | """Serialize an ``AnalysisReport`` into a frontend-ready JSON dict.
Everything a UI needs to render the verdict and every chart — forest plot,
score-distribution histograms, Bayesian posterior curve, mSPRT trajectory — is
pre-computed here so the client needs no numerical libraries.
"""
from __future__ import annotations
import numpy as np
def _f(x) -> float | None:
"""Cast to a JSON-safe float, mapping NaN/None to null."""
try:
if x is None:
return None
xf = float(x)
return None if np.isnan(xf) else xf
except (TypeError, ValueError):
return None
def report_to_dict(report, scoring, resume_a=None, resume_b=None) -> dict:
v = report.verdict
boot = report.bootstrap
clusters = []
for _, r in report.per_cluster.iterrows():
clusters.append(
{
"cluster_id": int(r["cluster_id"]),
"label": str(r["label"]),
"n": int(r["n"]),
"mean_delta": _f(r["mean_delta"]),
"ci_low": _f(r["ci_low"]),
"ci_high": _f(r["ci_high"]),
"p_raw": _f(r["p_raw"]),
"p_bonferroni": _f(r["p_bonferroni"]),
"p_bh_fdr": _f(r["p_bh_fdr"]),
"sig_bonferroni": bool(r["sig_bonferroni"]),
"sig_bh": bool(r["sig_bh"]),
"winner": str(r["winner"]),
"gaps": report.gaps.get(int(r["cluster_id"]), []),
"differentiators": report.differentiators.get(int(r["cluster_id"]), {}),
}
)
lo = float(min(scoring.scores_a.min(), scoring.scores_b.min()))
hi = float(max(scoring.scores_a.max(), scoring.scores_b.max()))
edges = np.linspace(lo, hi, 41)
a_counts, _ = np.histogram(scoring.scores_a, bins=edges)
b_counts, _ = np.histogram(scoring.scores_b, bins=edges)
centers = (edges[:-1] + edges[1:]) / 2
xs, pdf = report.bayes.posterior_pdf(128)
traj = report.sequential.trajectory_p
idx = np.unique(np.linspace(0, len(traj) - 1, 200).astype(int))
sequential_points = [{"n": int(i + 1), "p": float(traj[i])} for i in idx]
out = {
"verdict": {
"winner": v.winner,
"headline": v.headline,
"confidence": v.confidence,
"significant": bool(v.significant),
"mean_delta": _f(v.mean_delta),
"mean_delta_points": _f(v.mean_delta * 100),
"ci_points": [_f(boot.bca_low * 100), _f(boot.bca_high * 100)],
"p_value": _f(v.p_value),
"cohens_d": _f(v.cohens_d),
},
"summary": {
**{k: _f(val) for k, val in report.scores_summary.items()},
"n_jobs": report.n_jobs,
},
"tests": {
"primary": {
"name": report.primary_test.name,
"statistic": _f(report.primary_test.statistic),
"pvalue": _f(report.primary_test.pvalue),
"ci_low": _f(report.primary_test.ci_low),
"ci_high": _f(report.primary_test.ci_high),
},
"normality": {
"name": report.normality.name,
"statistic": _f(report.normality.statistic),
"pvalue": _f(report.normality.pvalue),
"normal_at_05": bool(report.normality.detail.get("normal_at_05", True)),
},
"cuped_test": {
"name": report.cuped_test.name,
"pvalue": _f(report.cuped_test.pvalue),
"ci_low": _f(report.cuped_test.ci_low),
"ci_high": _f(report.cuped_test.ci_high),
},
},
"effect": {
"cohens_d": _f(report.cohens_d),
"achieved_power": _f(report.achieved_power),
"required_n_80": _f(report.required_n_80),
"mde": report.mde.to_dict(orient="records"),
},
"bootstrap": {
"point": _f(boot.point),
"percentile": [_f(boot.pct_low), _f(boot.pct_high)],
"bca": [_f(boot.bca_low), _f(boot.bca_high)],
"n_resamples": boot.n_resamples,
},
"cuped": {
"variance_reduction": _f(report.cuped.variance_reduction),
"r_squared": _f(report.cuped.r_squared),
"effective_n_multiplier": _f(report.cuped.effective_n_multiplier),
"n_covariates": report.cuped.n_covariates,
},
"sequential": {
"always_valid_p": _f(report.sequential.always_valid_p),
"reject_h0": bool(report.sequential.reject_h0),
"n": report.sequential.n,
"trajectory": sequential_points,
},
"bayes": {
"k": report.bayes.k,
"n": report.bayes.n,
"posterior_mean": _f(report.bayes.mean),
"credible_interval": [_f(report.bayes.ci_low), _f(report.bayes.ci_high)],
"prob_b_beats_a": _f(report.bayes.prob_b_beats_a),
"posterior_curve": [
{"p": float(x), "density": float(y)} for x, y in zip(xs, pdf, strict=True)
],
},
"distributions": {
"bin_centers": [float(c) for c in centers],
"resume_a": [int(c) for c in a_counts],
"resume_b": [int(c) for c in b_counts],
},
"clusters": clusters,
}
if resume_a is not None and resume_b is not None:
out["inputs"] = {
slot: {
"chars": r.char_count,
"format": r.source_format,
"parser": r.parser_used,
"skills": r.skills,
"parse_quality": r.parse_quality,
"flags": r.quality_flags,
}
for slot, r in (("resume_a", resume_a), ("resume_b", resume_b))
}
return out
|