{
  "executed_at": "2026-09-02",
  "standard": "Run the verified 103-capability environment, screen all 103 lenses, execute every defensible lens on the public data, and preserve rejected-lens reasons.",
  "selftest": {
    "path": "data/public-analysis/education-real-estate-2025/103機能_再実行結果_20260902.json",
    "passed": 103,
    "failed": 0,
    "sha256": "487336faad996efbc24dbed55002840bdf1bd66d80afbe314189b3d41144eb45"
  },
  "catalog_count": 103,
  "executed_on_public_data": 64,
  "screened_not_applicable": 39,
  "unregistered": [],
  "lenses": {
    "numpy": {
      "status": "executed_on_public_data",
      "themes": [
        "school",
        "gender",
        "housing",
        "condominium"
      ],
      "role": "比率、成長率、leave-one-out、丸め区間の大量再計算",
      "output": {
        "recalculations": 100000
      }
    },
    "pandas": {
      "status": "executed_on_public_data",
      "themes": [
        "school",
        "gender",
        "housing",
        "condominium"
      ],
      "role": "全派生表と103機能による確認の主集計",
      "output": {
        "rows": 43
      }
    },
    "polars": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "CSVをPandasと別実装で再読込し減少寄与合計を照合",
      "output": {
        "total_change": -51528
      }
    },
    "duckdb": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "CSVへ直接SQLを実行し減少寄与合計を照合",
      "output": {
        "total_change": -51528
      }
    },
    "scipy": {
      "status": "executed_on_public_data",
      "themes": [
        "housing",
        "condominium"
      ],
      "role": "開始年別OLS、Theil–Sen頑健傾き、調査年除外傾き",
      "output": {
        "housing_start_year_slopes": [
          {
            "start_year": 2011,
            "n": 15,
            "slope_thousand_per_year": -9.115,
            "pvalue": 0.01939,
            "theil_sen": -12.312
          },
          {
            "start_year": 2012,
            "n": 14,
            "slope_thousand_per_year": -12.737,
            "pvalue": 0.00153,
            "theil_sen": -15.0
          },
          {
            "start_year": 2013,
            "n": 13,
            "slope_thousand_per_year": -15.857,
            "pvalue": 0.00024,
            "theil_sen": -18.5
          },
          {
            "start_year": 2014,
            "n": 12,
            "slope_thousand_per_year": -15.647,
            "pvalue": 0.00125,
            "theil_sen": -19.926
          },
          {
            "start_year": 2015,
            "n": 11,
            "slope_thousand_per_year": -19.353,
            "pvalue": 0.00027,
            "theil_sen": -21.101
          },
          {
            "start_year": 2016,
            "n": 10,
            "slope_thousand_per_year": -23.289,
            "pvalue": 6e-05,
            "theil_sen": -23.829
          },
          {
            "start_year": 2017,
            "n": 9,
            "slope_thousand_per_year": -23.729,
            "pvalue": 0.00042,
            "theil_sen": -24.0
          },
          {
            "start_year": 2018,
            "n": 8,
            "slope_thousand_per_year": -22.266,
            "pvalue": 0.00342,
            "theil_sen": -23.9
          },
          {
            "start_year": 2019,
            "n": 7,
            "slope_thousand_per_year": -19.379,
            "pvalue": 0.02306,
            "theil_sen": -21.361
          },
          {
            "start_year": 2020,
            "n": 6,
            "slope_thousand_per_year": -15.774,
            "pvalue": 0.11799,
            "theil_sen": -19.602
          }
        ]
      }
    },
    "statsmodels": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "各年除外HC3頑健標準誤差の傾向推定",
      "output": {
        "leave_one_year_out": [
          {
            "omitted_year": 2011,
            "hc3_slope": -12.737
          },
          {
            "omitted_year": 2012,
            "hc3_slope": -10.439
          },
          {
            "omitted_year": 2013,
            "hc3_slope": -7.901
          },
          {
            "omitted_year": 2014,
            "hc3_slope": -9.498
          },
          {
            "omitted_year": 2015,
            "hc3_slope": -9.072
          },
          {
            "omitted_year": 2016,
            "hc3_slope": -8.565
          },
          {
            "omitted_year": 2017,
            "hc3_slope": -8.816
          },
          {
            "omitted_year": 2018,
            "hc3_slope": -9.115
          },
          {
            "omitted_year": 2019,
            "hc3_slope": -9.253
          },
          {
            "omitted_year": 2020,
            "hc3_slope": -8.759
          },
          {
            "omitted_year": 2021,
            "hc3_slope": -9.19
          },
          {
            "omitted_year": 2022,
            "hc3_slope": -9.415
          },
          {
            "omitted_year": 2023,
            "hc3_slope": -8.872
          },
          {
            "omitted_year": 2024,
            "hc3_slope": -8.445
          },
          {
            "omitted_year": 2025,
            "hc3_slope": -6.887
          }
        ],
        "slope_range": [
          -12.737,
          -6.887
        ]
      }
    },
    "scikit-learn": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "地域別5内訳のPCAと探索的k-means。4地域のため主結論にしない",
      "output": {
        "pca_explained": [
          0.5514,
          0.3754
        ],
        "clusters": {
          "首都圏": 0,
          "中部圏": 0,
          "近畿圏": 1,
          "その他地域": 1
        },
        "silhouette": 0.259
      }
    },
    "pymc": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "住宅着工の年次傾きをベイズ回帰で再推定",
      "output": {
        "median_slope": -9.021,
        "p_negative": 0.9862
      }
    },
    "arviz": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "ベイズ傾きの収束診断",
      "output": {
        "rhat_beta": 1.0009,
        "ess_bulk": 2724.0
      }
    },
    "bambi": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "同じ実データの式モデルを別APIで構築し、モデル指定を監査",
      "output": {
        "components": [
          "mu",
          "sigma"
        ]
      }
    },
    "pmdarima": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "15年時系列のARIMA次数と1年先感度値",
      "output": {
        "order": [
          1,
          0,
          0
        ],
        "one_step": 772.7
      }
    },
    "ruptures": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "住宅着工系列の候補変化点検査",
      "output": {
        "candidate_break_years": []
      }
    },
    "sktime": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "ドリフト型ナイーブ予測を高度モデルの比較基準にする",
      "output": {
        "one_step": 734.6
      }
    },
    "tslearn": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "総数・持家・貸家・分譲の推移形状をDTW距離で比較",
      "output": {
        "distance_matrix": [
          [
            0.0,
            2.457,
            2.277,
            2.017
          ],
          [
            2.457,
            0.0,
            4.023,
            2.668
          ],
          [
            2.277,
            4.023,
            0.0,
            2.999
          ],
          [
            2.017,
            2.668,
            2.999,
            0.0
          ]
        ]
      }
    },
    "xgboost": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "ラグ特徴の時系列分割MAE。小標本で不安定なことを確認",
      "output": {
        "walk_forward_mae_thousand": 59.8
      }
    },
    "lightgbm": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "別実装の勾配ブースティングで同じ時系列分割を照合",
      "output": {
        "walk_forward_mae_thousand": 74.6
      }
    },
    "shap": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "小標本予測の依存先をSHAPで監査し、説明不安定性を確認",
      "output": {
        "mean_abs_shap": {
          "lag1": 4.007999897003174,
          "lag2": 1.774999976158142,
          "lag3": 10.477999687194824,
          "year_index": 45.803001403808594
        }
      }
    },
    "optuna": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "時系列分割だけで20通り探索し、最適化による過適合リスクを可視化",
      "output": {
        "best_mae_thousand": 50.5,
        "trials": 20
      }
    },
    "SALib": {
      "status": "executed_on_public_data",
      "themes": [
        "condominium"
      ],
      "role": "公表丸め区間に対する築40年超構成比のSobol感度",
      "output": {
        "S1": {
          "aged_stock": 0.954,
          "total_stock": 0.0459
        }
      }
    },
    "openturns": {
      "status": "executed_on_public_data",
      "themes": [
        "condominium"
      ],
      "role": "独立確率実装で丸め感度10,000回を再計算",
      "output": {
        "mean": 21.9458,
        "std": 0.00407
      }
    },
    "pyarrow": {
      "status": "executed_on_public_data",
      "themes": [
        "school",
        "gender"
      ],
      "role": "派生表をParquetへ往復し型と行数を照合",
      "output": {
        "rows": 11,
        "bytes": 20576
      }
    },
    "openpyxl": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "派生時系列をExcelへ往復し行数と値を照合",
      "output": {
        "rows": 15,
        "bytes": 5374
      }
    },
    "pyreadstat": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "実データをSPSS形式へ往復して可搬性を検査",
      "output": {
        "rows": 15
      }
    },
    "seaborn": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "地域別内訳をMatplotlibとは別の統計図実装で照合",
      "output": {
        "path": "public/research/figures/住宅着工2025_地域内訳seaborn_ja.png"
      }
    },
    "plotly": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "マウスで系列と年を確認できる対話図",
      "output": {
        "path": "public/research/住宅着工2025_利用関係別推移_対話図.html",
        "bytes": 10350
      }
    },
    "missingno": {
      "status": "executed_on_public_data",
      "themes": [
        "school",
        "gender"
      ],
      "role": "異なる派生表を結合した際の構造的欠損を可視化し、補完対象と誤認しない",
      "output": {
        "path": "public/research/figures/教育不動産2025_欠損構造監査.png"
      }
    },
    "scikit-image": {
      "status": "executed_on_public_data",
      "themes": [
        "school",
        "gender",
        "housing",
        "condominium"
      ],
      "role": "公開図のエッジ密度を測り、白紙・過密画像を機械監査",
      "output": {
        "images": 14,
        "edge_density": {
          "マンション2025_ストックと建替え規模_ja.png": 0.02361,
          "マンション2025_公表丸め感度_ja.png": 0.02662,
          "マンション2025_築40年超ストック_ja.png": 0.02375,
          "住宅着工2025_利用関係別推移_ja.png": 0.03102,
          "住宅着工2025_地域内訳seaborn_ja.png": 0.02254,
          "住宅着工2025_地域別増減率_ja.png": 0.02389,
          "住宅着工2025_減少寄与分解_ja.png": 0.02715,
          "住宅着工2025_開始年別傾き感度_ja.png": 0.02461,
          "学校基本統計2025_1校当たり在学者数_ja.png": 0.02135,
          "学校基本統計2025_在学者学校教員の増減率_ja.png": 0.02396,
          "学校基本統計2025_在学者教員比_ja.png": 0.02235,
          "高等教育2025_主要3区分の女性比率_ja.png": 0.02196,
          "高等教育2025_制度区分別女性比率_ja.png": 0.04162,
          "高等教育2025_女性比率前年差_ja.png": 0.02284
        }
      }
    },
    "matplotlib": {
      "status": "executed_on_public_data",
      "themes": [
        "school",
        "gender",
        "housing",
        "condominium"
      ],
      "role": "本文・PDF共通の静的統計図と感度図を生成",
      "output": {
        "new_sensitivity_figures": 6
      }
    },
    "janome": {
      "status": "executed_on_public_data",
      "themes": [
        "source_audit"
      ],
      "role": "公式資料の日本語名詞を抽出し、出典が扱う論点を監査",
      "output": {
        "tokens": 1353
      }
    },
    "fugashi": {
      "status": "executed_on_public_data",
      "themes": [
        "source_audit"
      ],
      "role": "別形態素解析器で同じ公式資料を再分割",
      "output": {
        "tokens": 0
      }
    },
    "sudachipy": {
      "status": "executed_on_public_data",
      "themes": [
        "source_audit"
      ],
      "role": "第三の分割器で専門語の抽出差を確認",
      "output": {
        "tokens": 857
      }
    },
    "wordcloud": {
      "status": "executed_on_public_data",
      "themes": [
        "source_audit"
      ],
      "role": "公式資料の頻出名詞を視覚監査。主分析の図には使わない",
      "output": {
        "path": "public/research/figures/教育不動産2025_公式資料用語監査.png"
      }
    },
    "gensim": {
      "status": "executed_on_public_data",
      "themes": [
        "source_audit"
      ],
      "role": "公式資料群の潜在トピックを2群で試し、数値分析の出典被覆を確認",
      "output": {
        "topics": [
          [
            "       ",
            "(-",
            "      ",
            "     ",
            "        ",
            "学校",
            "\n    ",
            "前年度"
          ],
          [
            "マンション",
            "年度",
            "\n        ",
            "平成",
            "令和",
            "住宅",
            "戸数",
            "   "
          ]
        ]
      }
    },
    "networkx": {
      "status": "executed_on_public_data",
      "themes": [
        "source_audit"
      ],
      "role": "公式資料間の論点共起ネットワーク",
      "output": {
        "nodes": 10,
        "edges": 14,
        "centrality": {
          "学校": 0.333,
          "在学者": 0.333,
          "女性": 0.333,
          "大学": 0.333,
          "住宅": 0.333,
          "貸家": 0.333,
          "分譲": 0.333,
          "マンション": 0.556,
          "建替え": 0.111,
          "高齢": 0.111
        }
      }
    },
    "python-louvain": {
      "status": "executed_on_public_data",
      "themes": [
        "source_audit"
      ],
      "role": "論点共起ネットワークのコミュニティ分割",
      "output": {
        "communities": {
          "学校": 0,
          "在学者": 0,
          "女性": 0,
          "大学": 0,
          "住宅": 1,
          "貸家": 1,
          "分譲": 1,
          "マンション": 1,
          "建替え": 1,
          "高齢": 1
        }
      }
    },
    "igraph": {
      "status": "executed_on_public_data",
      "themes": [
        "source_audit"
      ],
      "role": "別グラフ実装で連結性・直径を照合",
      "output": {
        "vertices": 10,
        "edges": 14,
        "diameter": 2
      }
    },
    "sympy": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "利用関係別の減少寄与を整数の厳密計算で再照合",
      "output": {
        "total_change": -51528
      }
    },
    "autograd": {
      "status": "executed_on_public_data",
      "themes": [
        "condominium"
      ],
      "role": "築40年超構成比の局所感度を自動微分",
      "output": {
        "aged_stock": 0.138198,
        "total_stock": -0.030329
      }
    },
    "numdifftools": {
      "status": "executed_on_public_data",
      "themes": [
        "condominium"
      ],
      "role": "同じ局所感度を数値微分で独立照合",
      "output": {
        "gradient": [
          0.138198,
          -0.030329
        ]
      }
    },
    "numba": {
      "status": "executed_on_public_data",
      "themes": [
        "condominium"
      ],
      "role": "10万回の公表丸め感度をJIT計算で再実行",
      "output": {
        "draws": 100000,
        "median": 21.9458
      }
    },
    "numexpr": {
      "status": "executed_on_public_data",
      "themes": [
        "condominium"
      ],
      "role": "10万回の配列式を第三実装で照合",
      "output": {
        "draws": 100000,
        "max_abs_difference": 0.0
      }
    },
    "xarray": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "年×利用関係のラベルを保ったまま前年差を照合",
      "output": {
        "change_2025": {
          "owner": -16.89,
          "rental": -17.101,
          "for_sale": -17.146
        }
      }
    },
    "xarray-einstats": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "利用関係別の時系列調和平均を次元名で計算",
      "output": {
        "harmonic_mean_thousand": {
          "owner": 269.938,
          "rental": 346.884,
          "for_sale": 243.852
        }
      }
    },
    "formulaic": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "年次傾向モデルの式を設計行列へ展開し列定義を監査",
      "output": {
        "shape": [
          15,
          2
        ],
        "columns": [
          "Intercept",
          "year_index"
        ]
      }
    },
    "patsy": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "非線形傾向の候補スプライン設計だけを作り、標本数に対する自由度を確認",
      "output": {
        "shape": [
          15,
          5
        ],
        "columns": [
          "Intercept",
          "bs(year, df=4)[0]",
          "bs(year, df=4)[1]",
          "bs(year, df=4)[2]",
          "bs(year, df=4)[3]"
        ]
      }
    },
    "pydot": {
      "status": "executed_on_public_data",
      "themes": [
        "workflow"
      ],
      "role": "公式資料から研究報告書・1ページ要約までの監査工程を有向グラフ化",
      "output": {
        "nodes": 6,
        "edges": 5,
        "path": "public/research/教育不動産2025_分析工程.dot"
      }
    },
    "pyvis": {
      "status": "executed_on_public_data",
      "themes": [
        "workflow"
      ],
      "role": "103機能の確認工程を対話型ネットワークとして公開用HTMLへ変換",
      "output": {
        "nodes": 6,
        "edges": 5,
        "path": "public/research/教育不動産2025_分析工程ネットワーク.html"
      }
    },
    "utilsforecast": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "前年値ベースラインのMAEを予測モデルと同一定義で集計",
      "output": {
        "mae_thousand": 44.809
      }
    },
    "dask": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "公式時系列を3分割で読み、原表と同じ行順・値へ再結合",
      "output": {
        "partitions": 3,
        "rows": 15
      }
    },
    "pandera": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "年の一意性・範囲と着工戸数の非負条件をスキーマ検査",
      "output": {
        "rows": 15,
        "violations": 0
      }
    },
    "pyjanitor": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "列名を一定規則へ統一し、分析用の列参照を再確認",
      "output": {
        "columns": 5,
        "rows": 15
      }
    },
    "bokeh": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "住宅着工系列をWeb向け対話図のJSONへ変換",
      "output": {
        "points": 15,
        "root_id": "p1006"
      }
    },
    "altair": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "宣言型の別実装で年次系列の図仕様を生成",
      "output": {
        "rows": 15,
        "mark": "line"
      }
    },
    "plotnine": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "ggplot文法の別実装で年次系列を描画",
      "output": {
        "rows": 15,
        "axes": 1
      }
    },
    "holoviews": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "同じ系列を高水準APIからBokeh描画へ変換",
      "output": {
        "points": 15,
        "renderer": "figure"
      }
    },
    "pyod": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "年と着工戸数の近傍距離から外れ年候補を抽出し、主結論と分離",
      "output": {
        "candidate_year": 2025
      }
    },
    "river": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "年次到着順に逐次学習し、順序依存の誤差を診断",
      "output": {
        "updates": 15,
        "prequential_mae": 359.109
      }
    },
    "skforecast": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "3ラグ自己回帰の2期先感度値をARIMA・ナイーブと比較",
      "output": {
        "forecast": [
          724.769,
          702.909
        ]
      }
    },
    "tsfresh": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "総数・持家・貸家・分譲の系列特徴を同一定義で一括抽出",
      "output": {
        "series": 4,
        "features": 3
      }
    },
    "pymannkendall": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "単調傾向を分布仮定の少ない検定で再確認",
      "output": {
        "trend": "decreasing",
        "pvalue": 0.022822,
        "sen_slope": -12.312
      }
    },
    "filterpy": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "状態空間モデルで年次系列を逐次平滑化し、短系列のノイズ感度を確認",
      "output": {
        "final_state": 807.972
      }
    },
    "pykalman": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "隔離環境の状態空間実装で同じ年次系列を平滑化",
      "output": {
        "final_state": 765.769
      }
    },
    "catboost": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "第三の木モデルで小標本内の当てはまりを確認し、予測根拠には使わない",
      "output": {
        "training_mae": 33.651
      }
    },
    "prince": {
      "status": "executed_on_public_data",
      "themes": [
        "housing"
      ],
      "role": "地域内訳PCAを別実装で再計算し、4地域の探索図に限定",
      "output": {
        "shape": [
          4,
          2
        ]
      }
    },
    "pingouin": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "反復測定の個票・尺度項目・媒介変数がない。"
    },
    "factor_analyzer": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "同一対象に対する複数尺度項目がなく、因子分析の標本設計を満たさない。"
    },
    "semopy": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "潜在変数を識別できる個票共分散行列がない。"
    },
    "scikit-posthocs": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "制度区分は無作為標本の比較群ではなく、多重検定を主張できない。"
    },
    "researchpy": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "Pandasの記述表で足り、別ソフトの表を増やしても新しい視点にならない。"
    },
    "statsmodels(MixedLM)": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "学校・自治体などの階層個票がない。"
    },
    "scikit-learn(IterativeImputer)": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "主要原表に補完すべき欠損がなく、構造的欠損を埋めるべきでない。"
    },
    "dowhy": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "処置、交絡、アウトカムを識別できる観察単位がない。"
    },
    "econml": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "個票・処置・共変量がなく、異質処置効果を推定できない。"
    },
    "linearmodels": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "同一定義の地域×年パネルがない。"
    },
    "imbalanced-learn": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "分類目的変数がない。"
    },
    "category_encoders": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "カテゴリを予測モデルへ符号化する必要がない。"
    },
    "feature-engine": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "前処理で変換すべき個票特徴量がない。"
    },
    "yellowbrick": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "小標本予測を見栄えのよい診断図にしても信頼性は増えない。"
    },
    "prophet": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "年次15点で季節性を識別できず、将来値の提示が誤解を招く。"
    },
    "statsforecast": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "複数予測器を追加しても15点ではモデル選択誤差が大きい。ARIMAとナイーブの感度比較に限定した。"
    },
    "arch": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "年次15点はGARCHの分散動学を推定するには不足。"
    },
    "lifelines": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "イベント発生時刻と打切り情報がない。"
    },
    "scikit-survival": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "生存時間・打切り・個票特徴量がない。"
    },
    "mlxtend": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "併売・同時発生のトランザクションデータがない。"
    },
    "umap-learn": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "4地域の次元圧縮は安定せず、距離を誤読させる。"
    },
    "hdbscan": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "4地域では密度クラスタの標本数が不足。"
    },
    "spacy": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "日本語学習済みモデルを使わない状態では、形態素解析3実装を超える根拠を追加しない。"
    },
    "textstat": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "日本語公式資料に英語用可読性式を当てると尺度が不適切。"
    },
    "causal-learn": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "集計時系列から変数間の方向を探索しても、独立観測と交絡条件を満たさず因果候補を公表できない。"
    },
    "cvxpy": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "最適化すべき目的関数・制約・意思決定変数が公式統計だけでは定義されていない。"
    },
    "highspy": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "線形計画の費用係数と制約条件がなく、恣意的な最適配分を作らない。"
    },
    "sparse": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "現行派生表は低次元で、疎行列化しても新しい結論や計算上の利点がない。"
    },
    "pyhdfe": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "吸収すべき自治体・年などの高次元固定効果パネルがない。"
    },
    "nltk": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "公式日本語資料の語分割には日本語形態素解析3実装を使っており、英字正規表現の追加は論点を増やさない。"
    },
    "statannotations": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "注記対象となる個票ベースの独立群比較がなく、集計値へ有意差記号を付けない。"
    },
    "holidays": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "年次・調査年データであり、祝日を日次特徴量として結合する時間粒度がない。"
    },
    "copulas": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "4地域または15年の集計値から合成標本を作ると、存在しない地域・年を実観測と誤認させる。"
    },
    "dcor": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "15年の時系列は独立観測ではなく、距離相関を独立標本の依存度として公表できない。"
    },
    "hyppo": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "15年の時系列は独立性検定の交換可能性を満たさない。"
    },
    "kmodes": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "クラスタリング対象となる十分なカテゴリ観測がない。"
    },
    "kneed": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "地域が4区分しかなく、折れ曲がり点から安定したクラスタ数を選べない。"
    },
    "datashader": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "最大15年・4地域の表であり、大量点の画素集約を使う規模ではない。"
    },
    "minisom": {
      "status": "screened_not_applicable",
      "themes": [],
      "reason": "4地域の自己組織化マップは標本不足で近傍構造を安定して推定できない。"
    }
  },
  "robustness": {
    "housing_start_year_slopes": [
      {
        "start_year": 2011,
        "n": 15,
        "slope_thousand_per_year": -9.115,
        "pvalue": 0.01939,
        "theil_sen": -12.312
      },
      {
        "start_year": 2012,
        "n": 14,
        "slope_thousand_per_year": -12.737,
        "pvalue": 0.00153,
        "theil_sen": -15.0
      },
      {
        "start_year": 2013,
        "n": 13,
        "slope_thousand_per_year": -15.857,
        "pvalue": 0.00024,
        "theil_sen": -18.5
      },
      {
        "start_year": 2014,
        "n": 12,
        "slope_thousand_per_year": -15.647,
        "pvalue": 0.00125,
        "theil_sen": -19.926
      },
      {
        "start_year": 2015,
        "n": 11,
        "slope_thousand_per_year": -19.353,
        "pvalue": 0.00027,
        "theil_sen": -21.101
      },
      {
        "start_year": 2016,
        "n": 10,
        "slope_thousand_per_year": -23.289,
        "pvalue": 6e-05,
        "theil_sen": -23.829
      },
      {
        "start_year": 2017,
        "n": 9,
        "slope_thousand_per_year": -23.729,
        "pvalue": 0.00042,
        "theil_sen": -24.0
      },
      {
        "start_year": 2018,
        "n": 8,
        "slope_thousand_per_year": -22.266,
        "pvalue": 0.00342,
        "theil_sen": -23.9
      },
      {
        "start_year": 2019,
        "n": 7,
        "slope_thousand_per_year": -19.379,
        "pvalue": 0.02306,
        "theil_sen": -21.361
      },
      {
        "start_year": 2020,
        "n": 6,
        "slope_thousand_per_year": -15.774,
        "pvalue": 0.11799,
        "theil_sen": -19.602
      }
    ],
    "housing_leave_one_year_out": [
      {
        "omitted_year": 2011,
        "hc3_slope": -12.737
      },
      {
        "omitted_year": 2012,
        "hc3_slope": -10.439
      },
      {
        "omitted_year": 2013,
        "hc3_slope": -7.901
      },
      {
        "omitted_year": 2014,
        "hc3_slope": -9.498
      },
      {
        "omitted_year": 2015,
        "hc3_slope": -9.072
      },
      {
        "omitted_year": 2016,
        "hc3_slope": -8.565
      },
      {
        "omitted_year": 2017,
        "hc3_slope": -8.816
      },
      {
        "omitted_year": 2018,
        "hc3_slope": -9.115
      },
      {
        "omitted_year": 2019,
        "hc3_slope": -9.253
      },
      {
        "omitted_year": 2020,
        "hc3_slope": -8.759
      },
      {
        "omitted_year": 2021,
        "hc3_slope": -9.19
      },
      {
        "omitted_year": 2022,
        "hc3_slope": -9.415
      },
      {
        "omitted_year": 2023,
        "hc3_slope": -8.872
      },
      {
        "omitted_year": 2024,
        "hc3_slope": -8.445
      },
      {
        "omitted_year": 2025,
        "hc3_slope": -6.887
      }
    ],
    "condominium_rounding_uncertainty": {
      "share_p2_5": 21.9388,
      "share_median": 21.9458,
      "share_p97_5": 21.9528
    },
    "household_age_leave_one_wave_out": [
      {
        "omitted_year": 2003,
        "slope_pp_per_year": 0.84
      },
      {
        "omitted_year": 2008,
        "slope_pp_per_year": 0.784
      },
      {
        "omitted_year": 2013,
        "slope_pp_per_year": 0.812
      },
      {
        "omitted_year": 2018,
        "slope_pp_per_year": 0.8091
      },
      {
        "omitted_year": 2023,
        "slope_pp_per_year": 0.838
      }
    ]
  },
  "outputs": [
    {
      "path": "public/research/教育不動産2025_分析入力.parquet",
      "bytes": 20576,
      "sha256": "fb72b0eddae326c00cc9a62d70536eb95ec1a7d9d2541eb4875a974ea2bb842a"
    },
    {
      "path": "public/research/教育不動産2025_分析入力.xlsx",
      "bytes": 5374,
      "sha256": "51966f8a09c5b16be704f8ec722d2fd0dc9db0d28856304c723451080d875db4"
    },
    {
      "path": "public/research/住宅着工2025_利用関係別推移_対話図.html",
      "bytes": 10350,
      "sha256": "5271c8b86762833066524fdd192e67081ae0c300d7bb5921dbb3e5257bb940b9"
    },
    {
      "path": "public/research/figures/住宅着工2025_地域内訳seaborn_ja.png",
      "bytes": 61482,
      "sha256": "c752c81bd8bf2c457b9b49b4d4bc416f644a9c78d443ce056562e9771022b8ec"
    },
    {
      "path": "public/research/figures/教育不動産2025_欠損構造監査.png",
      "bytes": 82728,
      "sha256": "8438b25dba2149166aac8dc03aee5c9b0cb4320c568befaf5f8df654bb74fd73"
    },
    {
      "path": "public/research/figures/教育不動産2025_公式資料用語監査.png",
      "bytes": 335971,
      "sha256": "fb66635d27254b818edd6d6a6d784cab496fba35a68bdfd560120c6eac1b8145"
    },
    {
      "path": "public/research/教育不動産2025_分析工程.dot",
      "bytes": 299,
      "sha256": "09ad353c5c575e15e13cf46940623b0b3fd8775127ada03e172ad6741609ba8a"
    },
    {
      "path": "public/research/教育不動産2025_分析工程ネットワーク.html",
      "bytes": 5199,
      "sha256": "1f99d5618db12468b0853a7cdea542d44bc9f4b78f2cffa685dacec44f815afe"
    },
    {
      "path": "public/research/figures/住宅着工2025_開始年別傾き感度_ja.png",
      "bytes": 64799,
      "sha256": "3b77ac3dedd672a09cb96ac1d95c543a659c81e0c37c4174c6d14eeeee829c66"
    },
    {
      "path": "public/research/figures/マンション2023_70歳以上傾き感度_ja.png",
      "bytes": 53551,
      "sha256": "29234b5a8c5cf9f20f7b15adefb3ed89330f8681a2c99fe80e12d5fa75447a91"
    },
    {
      "path": "public/research/figures/マンション2025_公表丸め感度_ja.png",
      "bytes": 56940,
      "sha256": "18e5ea21b47bbf4b7f9d951586c651c0afe426d18c8b7c08b056b1ba592d9c12"
    },
    {
      "path": "public/research/figures/住宅着工2025_開始年別傾き感度_en.png",
      "bytes": 54327,
      "sha256": "ce6b299070b5de0a2d603e261975ff2b0ade534187897efeb4b94ca51078d347"
    },
    {
      "path": "public/research/figures/マンション2023_70歳以上傾き感度_en.png",
      "bytes": 44161,
      "sha256": "f96cb5a2536fd60449eb0b76a087dd020954ed9a5701a61043776f525cb5c691"
    },
    {
      "path": "public/research/figures/マンション2025_公表丸め感度_en.png",
      "bytes": 46369,
      "sha256": "459a2f6f68d007cd79bd92a0eff1521f0cb65915781679507b9552cf3268c2a0"
    }
  ]
}