{"record": {"aliases": ["synthetic-data-benchmark"], "artifacts": [{"id": "gh:mervecaliskann/synthetic-data-benchmark", "kind": "repo", "url": "https://github.com/Mervecaliskann/synthetic-data-benchmark"}], "categories": ["Finance", "Language & Knowledge", "cs.AI"], "description": {"en": "The project evaluates synthetic tabular data generators on fidelity, utility, and privacy metrics using the Give Me Some Credit dataset, comparing Gaussian Copula, CTGAN, and an LLM generator over a fixed evaluation protocol."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://github.com/Mervecaliskann/synthetic-data-benchmark", "source": "claire_radar", "source_url": "https://github.com/Mervecaliskann/synthetic-data-benchmark", "title": "Synthetic Data Generator Benchmark (Credit Risk)"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:github:mervecaliskann/synthetic-data-benchmark", "modality": null, "name": "Synthetic Data Generator Benchmark (Credit Risk)", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"admission_policy_version": "2026-09-09.1", "confidence": "Medium", "crawl_bundle": "claire_radar_2026-10-14", "crawled_at": "2026-10-14T09:44:51Z", "data_status": "primary-source-candidate", "display_eligible": "false", "origin_record_id": "bm_synthetic-data-benchmark_e9558d16", "origin_source": "github", "recognition_confidence": "0.4", "record_sha256": "88a8f8a8643ccd967b0a4277c9809465b169edd17e38c1a8249d88838de2b8db", "relation": "unclear", "review_model": "deepseek-v4-pro", "review_state": "ai-name-audit-deferred", "reviewed_at": "2026-09-11T00:52:53.785364Z", "source_url": "https://github.com/Mervecaliskann/synthetic-data-benchmark"}, "publisher": null, "released": "2026-09-07", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:github:mervecaliskann/synthetic-data-benchmark", "source_url": "https://github.com/Mervecaliskann/synthetic-data-benchmark"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-github-mervecaliskann-synthetic-data-benchmark", "source": "claire_radar", "source_benchmark_id": "github:mervecaliskann/synthetic-data-benchmark", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": ["synthetic-data-benchmark"], "annotation": "Machine Generated", "applicationDomains": ["Finance"], "area": "Language & Knowledge", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": "https://github.com/Mervecaliskann/synthetic-data-benchmark", "githubScope": "benchmark_repo", "githubStars": 0, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": null, "hfPaperUrl": null, "observedAt": "2026-10-04T15:31:48.030916Z", "signalStatus": {"githubStars": {"observedAt": "2026-10-04T15:31:48.030916Z", "state": "fresh"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"state": "not_applicable"}}, "status": "fresh"}, "attentionForecast": {"confidence": "Low", "horizon": "7d", "reason": "The synthetic data benchmarking niche with focus on credit risk and a novel LLM generator may attract moderate interest from practitioners seeking reproducible comparisons, but limited initial visibility and single-run results temper expectations.", "score": 35}, "capabilities": [], "confidence": "Medium", "construction": "Original Synthetic", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-09-11T00:52:53.785364Z", "inputHash": "9f69bd5c55280af3a914710bf1674a213a507d871abfb79891c9fd6d9896bcb5", "model": "deepseek-v4-pro"}, "curation": {"admissionPolicyVersion": "2026-09-09.1", "canonicalNameEvidence": "# Synthetic Data Generator Benchmark (Credit Risk)", "canonicalNameSource": "official_readme", "decisionReason": "The benchmark results are based on a single run with n_repeats=1, and repeated runs with statistical significance testing are listed as planned next steps, so the scoring contract is not yet stable enough for ongoing comparisons.", "evaluationEvidence": {"reuse": "The leaderboard is written to `results/leaderboard.csv`, and per-repeat results are written to `results/raw_results.csv`. These results come from a single run with n_repeats set to 1.", "scoring": "Fidelity and utility are measured against the holdout set, since that data was never seen by any generator during fitting. Privacy is measured differently: distance to closest record is computed against the training set, because memorization risk concerns the data a generator actually trained on, not data it never saw.", "task": "Three generators are compared: Gaussian Copula, CTGAN (fit on a 25,000 row subsample for 100 epochs to keep CPU runtime reasonable), and an LLM generator built on Groq's openai/gpt-oss-120b model."}, "model": "deepseek-v4-pro", "reviewedAt": "2026-09-11T00:52:53.785364Z", "state": "ai-name-audit-deferred"}, "dataStatus": "primary-source-candidate", "demo": false, "description": "The project evaluates synthetic tabular data generators on fidelity, utility, and privacy metrics using the Give Me Some Credit dataset, comparing Gaussian Copula, CTGAN, and an LLM generator over a fixed evaluation protocol.", "discoverySources": [{"id": "github:mervecaliskann/synthetic-data-benchmark", "type": "github", "url": "https://github.com/Mervecaliskann/synthetic-data-benchmark"}], "displayEligible": false, "domainCuration": {"method": "rules-v1", "state": "auto"}, "evaluationMode": "score_submission", "evidence": {"reasonCodes": ["discovered via github", "benchmark term in abstract", "evaluation protocol evidence", "no explicit benchmark release evidence"], "snippet": "synthetic-data-benchmark Benchmarking synthetic tabular data generators (Gaussian Copula, CTGAN, LLM) on fidelity, utility, and privacy, using the Give Me Some Credit dataset."}, "familyId": "bmf_1ea7a753c9bf", "firstSeenAt": "2026-09-08", "heat": null, "id": "bm_synthetic-data-benchmark_e9558d16", "indexedAt": "2026-09-08T06:12:01.862698Z", "industrySectors": ["Financial Services"], "links": {"code": "https://github.com/Mervecaliskann/synthetic-data-benchmark"}, "metrics": [], "motivation": "synthetic-data-benchmark Benchmarking synthetic tabular data generators (Gaussian Copula, CTGAN, LLM) on fidelity, utility, and privacy, using the Give Me Some Credit dataset.", "name": "Synthetic Data Generator Benchmark (Credit Risk)", "oneLine": "The project evaluates synthetic tabular data generators on fidelity, utility, and privacy metrics using the Give Me Some Credit dataset, comparing Gaussian Copula, CTGAN, and an LLM generator over a fixed evaluation protocol.", "paperTitle": "synthetic-data-benchmark", "primaryDomain": "Finance", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://github.com/Mervecaliskann/synthetic-data-benchmark", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-09-08T06:12:01.862698Z"}, "publications": [], "ranking": {"30d": {"confidence": "Low", "coverage": 0.55, "growth": {"components": {"githubStars": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfDatasetDownloads": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfPaperUpvotes": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "elapsedDays": 30, "historyDate": "2026-09-05", "rank": null, "score": null, "windowDays": 30}, "level": {"components": {"githubStars": {"normalization": "log-max", "normalized": 0.0, "value": 0}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": null, "value": null}}, "confidence": "Low", "coverage": 0.55, "rank": 196, "score": 23}, "method": "attention-ranking-v13", "rank": 196, "score": 23}, "90d": {"confidence": "Low", "coverage": 0.55, "growth": {"components": {"githubStars": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfDatasetDownloads": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfPaperUpvotes": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "elapsedDays": null, "historyDate": null, "rank": null, "score": null, "windowDays": 90}, "level": {"components": {"githubStars": {"normalization": "log-max", "normalized": 0.0, "value": 0}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": null, "value": null}}, "confidence": "Low", "coverage": 0.55, "rank": 460, "score": 23}, "method": "attention-ranking-v13", "rank": 460, "score": 23}}, "readiness": "Runnable", "recognitionConfidence": 0.4, "relation": "unclear", "releasedAt": "2026-09-07", "source": {"authors": ["Mervecaliskann"], "categories": [], "id": "github:mervecaliskann/synthetic-data-benchmark", "publicSignals": {"githubStars": 0}, "title": "synthetic-data-benchmark", "type": "github", "url": "https://github.com/Mervecaliskann/synthetic-data-benchmark"}, "sourceUpdatedAt": "2026-09-07T15:55:39Z", "topics": ["cs.AI"], "venueAttempts": [], "whyItMatters": "It provides a reproducible methodology for comparing synthetic data generators, helping practitioners choose generators that best preserve distribution, downstream utility, and privacy for credit risk data."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
