Update app.py

app.py

@@ -4,25 +4,20 @@ IPLM 2025 — FINAL (NO UPLOAD) — FULL REWRITE (NO RINGKAS)
 
 ✅ Jenis tampil: sekolah, umum, khusus (khusus ditampilkan sebagai jenis)
 ✅ Indeks dasar per entitas: Yeo-Johnson + MinMax nasional per indikator
-✅ Penyesuaian 68% berbasis TOTAL pengumpulan wilayah:
+✅ Agregasi wilayah (mean) → penyesuaian 68% berbasis TOTAL pengumpulan wilayah:
    faktor_penyesuaian = min(n_total_terkumpul / target_total_68, 1.0)
-✅ KONSISTENSI UTAMA (FIX INKONSISTENSI):
-   Indeks_Dasar_Agregat_0_100 (KESELURUHAN wilayah) = (dasar_sekolah + dasar_umum + dasar_khusus) / 3 (missing=0, tetap ÷3)
-   Indeks_Final_Wilayah_0_100 = Indeks_Dasar_Agregat_0_100 × faktor_penyesuaian
-   -> Maka nilai Aceh Jaya di agregat provinsi akan sama dengan nilai Aceh Jaya saat dipilih individu (definisi keseluruhan sama).
-✅ Ringkasan (Jenis + Keseluruhan) selalu 4 baris: sekolah, umum, khusus, keseluruhan
-✅ Indeks_Final_Disesuaikan_0_100 (keseluruhan) = (final_sekolah+final_umum+final_khusus)/3 (missing=0, tetap ÷3)
-✅ Dashboard KPI FINAL mengambil dari Ringkasan (baris keseluruhan)
-✅ Detail entitas: Indeks_Final_0_100 menempel dari Indeks_Final_Wilayah_0_100 (bukan per-row)
+   Indeks_Final_Wilayah = Indeks_Dasar_Agregat * faktor_penyesuaian
+✅ Detail entitas: Indeks_Final_0_100 menempel dari Indeks_Final_Wilayah (bukan per-row)
 ✅ Bell curve per JENIS berbasis indeks per entitas (row-level)
 ✅ LLM analysis + Word
 ✅ Download (tanpa upload box)
-✅ Download Data Mentah (.xlsx) = RAW hasil filter (bukan agregat jenis)
-✅ Word report: hilangkan "Kewenangan: ..." ganti "Ringkasan Dashboard"
 
-CATATAN:
-- Kolom sub/dim pada agregat wilayah tetap dihitung mean seluruh entitas (itu tidak memengaruhi konsistensi indeks keseluruhan).
-- Konsistensi indeks keseluruhan dijamin karena semua level pakai definisi ÷3.
+PERBAIKAN UTAMA:
+✅ Ringkasan (Jenis + Keseluruhan) selalu tampil 4 baris: sekolah, umum, khusus, keseluruhan
+✅ Indeks_Final_Disesuaikan_0_100 (keseluruhan) = (final_sekolah+final_umum+final_khusus)/3 (missing=0, tetap ÷3)
+✅ Dashboard KPI FINAL mengambil dari Ringkasan (baris keseluruhan) — bukan mean baris yang tersedia
+✅ Word report: hilangkan "Kewenangan: ..." ganti "Ringkasan Dashboard" (mengacu KPI & Ringkasan)
+✅ Tombol "Download Data Mentah (.xlsx)" berisi RAW hasil filter (bukan agregat jenis)
 """
 
 import os
@@ -592,7 +587,6 @@ def load_default_files(force=False):
 
 # ============================================================
 # 6) AGREGAT WILAYAH (KESELURUHAN) + PENYESUAIAN TOTAL
-#    FIX INKONSISTENSI: "Indeks_Dasar_Agregat_0_100" pakai rata-rata 3 jenis ÷3
 # ============================================================
 
 def build_agg_wilayah_total(df_filtered: pd.DataFrame, pop_kab: pd.DataFrame, pop_prov: pd.DataFrame, kew_value: str):
@@ -627,8 +621,7 @@ def build_agg_wilayah_total(df_filtered: pd.DataFrame, pop_kab: pd.DataFrame, po
         pop_field = "Pop_Total"
         name_field = "Kab_Kota_Label"
 
-    # agregat mean sub/dim (boleh weighted by count, tidak untuk indeks keseluruhan)
-    agg_base = df.groupby([key_col, label_col], dropna=False).agg(
+    agg = df.groupby([key_col, label_col], dropna=False).agg(
         n_total=("Indeks_Dasar_0_100", "size"),
         Rata2_sub_koleksi=("sub_koleksi", "mean"),
         Rata2_sub_sdm=("sub_sdm", "mean"),
@@ -636,39 +629,11 @@ def build_agg_wilayah_total(df_filtered: pd.DataFrame, pop_kab: pd.DataFrame, po
         Rata2_sub_pengelolaan=("sub_pengelolaan", "mean"),
         Rata2_dim_kepatuhan=("dim_kepatuhan", "mean"),
         Rata2_dim_kinerja=("dim_kinerja", "mean"),
+        Indeks_Dasar_Agregat_0_100=("Indeks_Dasar_0_100", "mean"),
     ).reset_index()
 
-    # ===== KUNCI KONSISTENSI =====
-    # indeks dasar per jenis per wilayah, lalu overall = (sekolah+umum+khusus)/3 (missing=0)
-    jenis_mean = (
-        df[df["_dataset"].isin(["sekolah", "umum", "khusus"])]
-        .groupby([key_col, label_col, "_dataset"], dropna=False)["Indeks_Dasar_0_100"]
-        .mean()
-        .reset_index()
-    )
-
-    piv = jenis_mean.pivot_table(
-        index=[key_col, label_col],
-        columns="_dataset",
-        values="Indeks_Dasar_0_100",
-        aggfunc="mean"
-    ).reset_index()
-
-    for j in ["sekolah", "umum", "khusus"]:
-        if j not in piv.columns:
-            piv[j] = 0.0
-    piv[["sekolah", "umum", "khusus"]] = piv[["sekolah", "umum", "khusus"]].fillna(0.0)
-    piv["Indeks_Dasar_Agregat_0_100"] = (piv["sekolah"] + piv["umum"] + piv["khusus"]) / 3.0
-
-    agg = agg_base.merge(
-        piv[[key_col, label_col, "Indeks_Dasar_Agregat_0_100"]],
-        on=[key_col, label_col],
-        how="left"
-    )
-
     agg = agg.rename(columns={key_col: "group_key", label_col: label_name})
 
-    # attach target/pop
     target_vals, pop_vals, label_fix = [], [], []
     for _, r in agg.iterrows():
         gk = r["group_key"]
@@ -705,10 +670,8 @@ def build_agg_wilayah_total(df_filtered: pd.DataFrame, pop_kab: pd.DataFrame, po
         )
     ]
 
-    # FINAL wilayah konsisten
     agg["Indeks_Final_Wilayah_0_100"] = agg["Indeks_Dasar_Agregat_0_100"] * agg["faktor_penyesuaian"]
 
-    # rounding
     for c in [
         "Rata2_sub_koleksi","Rata2_sub_sdm","Rata2_sub_pelayanan","Rata2_sub_pengelolaan",
         "Rata2_dim_kepatuhan","Rata2_dim_kinerja"
@@ -777,12 +740,11 @@ def build_agg_wilayah_jenis(df_filtered: pd.DataFrame, agg_total: pd.DataFrame,
     agg["faktor_penyesuaian"] = agg["faktor_penyesuaian_wilayah"]
     agg["Indeks_Final_Agregat_0_100"] = agg["Indeks_Dasar_Agregat_0_100"] * agg["faktor_penyesuaian"]
 
-    # khusus: optional attach target/pop jenis (hanya untuk info)
     agg["target_total_68_jenis"] = np.nan
     agg["pop_total_jenis"] = np.nan
     agg["coverage_jenis"] = np.nan
 
-    if (pop_khusus is not None) and (not pop_khusus.empty) and ("KAB" in kew_norm or "KOTA" in kew_norm or kew_norm in {"(SEMUA)", "(SEMUA)".upper()}):
+    if (pop_khusus is not None) and (not pop_khusus.empty) and ("KAB" in kew_norm or "KOTA" in kew_norm or kew_norm == "(SEMUA)" or kew_norm == "(SEMUA)".upper()):
         pk = pop_khusus.set_index("kab_key")
         for i, r in agg.iterrows():
             if str(r.get("Jenis", "")).lower() != "khusus":
@@ -803,7 +765,6 @@ def build_agg_wilayah_jenis(df_filtered: pd.DataFrame, agg_total: pd.DataFrame,
     m2 = agg["pop_total_jenis"].notna() & (agg["pop_total_jenis"] > 0)
     agg.loc[m2, "coverage_jenis"] = (agg.loc[m2, "Jumlah"].astype(float) / agg.loc[m2, "pop_total_jenis"].astype(float)) * 100.0
 
-    # rounding
     for c in [
         "Rata2_sub_koleksi","Rata2_sub_sdm","Rata2_sub_pelayanan","Rata2_sub_pengelolaan",
         "Rata2_dim_kepatuhan","Rata2_dim_kinerja"
@@ -829,8 +790,10 @@ def build_agg_wilayah_jenis(df_filtered: pd.DataFrame, agg_total: pd.DataFrame,
 # ============================================================
 
 def build_summary_per_jenis(agg_jenis: pd.DataFrame, agg_total: pd.DataFrame):
+    # Wajib selalu ada 3 jenis, walau 0
     jenis_list = ["sekolah", "umum", "khusus"]
 
+    # helper bikin row default
     def _row_default(jenis):
         return {
             "Jenis": jenis,
@@ -852,6 +815,7 @@ def build_summary_per_jenis(agg_jenis: pd.DataFrame, agg_total: pd.DataFrame):
             sub = agg_jenis[agg_jenis["Jenis"].astype(str).str.lower() == jenis].copy()
             if sub.empty:
                 continue
+
             rows_by_jenis[jenis] = {
                 "Jenis": jenis,
                 "Jumlah_Wilayah": int(sub.shape[0]),
@@ -862,16 +826,20 @@ def build_summary_per_jenis(agg_jenis: pd.DataFrame, agg_total: pd.DataFrame):
                 "Rata2_sub_pengelolaan": float(pd.to_numeric(sub["Rata2_sub_pengelolaan"], errors="coerce").fillna(0).mean()),
                 "Rata2_dim_kepatuhan": float(pd.to_numeric(sub["Rata2_dim_kepatuhan"], errors="coerce").fillna(0).mean()),
                 "Rata2_dim_kinerja": float(pd.to_numeric(sub["Rata2_dim_kinerja"], errors="coerce").fillna(0).mean()),
+                # FINAL PER JENIS = mean Indeks_Final_Agregat_0_100 (wilayah×jenis)
                 "Indeks_Final_Disesuaikan_0_100": float(pd.to_numeric(sub["Indeks_Final_Agregat_0_100"], errors="coerce").fillna(0).mean()),
             }
 
+    # susun output 3 jenis
     rows = [rows_by_jenis[j] for j in jenis_list]
 
+    # KESELURUHAN harus ÷3 (missing=0)
     final_sekolah = float(rows_by_jenis["sekolah"]["Indeks_Final_Disesuaikan_0_100"])
     final_umum    = float(rows_by_jenis["umum"]["Indeks_Final_Disesuaikan_0_100"])
     final_khusus  = float(rows_by_jenis["khusus"]["Indeks_Final_Disesuaikan_0_100"])
     final_all = (final_sekolah + final_umum + final_khusus) / 3.0
 
+    # untuk indikator lain pada keseluruhan, juga pakai rata-rata 3 jenis (missing=0)
     def _avg3(field):
         return (float(rows_by_jenis["sekolah"][field]) + float(rows_by_jenis["umum"][field]) + float(rows_by_jenis["khusus"][field])) / 3.0
 
@@ -892,11 +860,12 @@ def build_summary_per_jenis(agg_jenis: pd.DataFrame, agg_total: pd.DataFrame):
         "Rata2_sub_pengelolaan": _avg3("Rata2_sub_pengelolaan"),
         "Rata2_dim_kepatuhan": _avg3("Rata2_dim_kepatuhan"),
         "Rata2_dim_kinerja": _avg3("Rata2_dim_kinerja"),
-        "Indeks_Final_Disesuaikan_0_100": final_all,
+        "Indeks_Final_Disesuaikan_0_100": final_all,  # FIX: selalu ÷3
     })
 
     out = pd.DataFrame(rows)
 
+    # rounding
     for c in [
         "Rata2_sub_koleksi","Rata2_sub_sdm","Rata2_sub_pelayanan","Rata2_sub_pengelolaan",
         "Rata2_dim_kepatuhan","Rata2_dim_kinerja"
@@ -1117,17 +1086,19 @@ def _make_bell_curve(dfp: pd.DataFrame, xcol: str, title: str, label_col: str |
 # ============================================================
 
 def compute_dashboard_kpis(summary_jenis: pd.DataFrame, agg_total: pd.DataFrame, agg_jenis: pd.DataFrame):
+    # pastikan ada baris 3 jenis + keseluruhan
     def _get_final(j):
         sub = summary_jenis[summary_jenis["Jenis"].astype(str).str.lower() == j]
         if sub.empty:
             return 0.0
         return float(pd.to_numeric(sub["Indeks_Final_Disesuaikan_0_100"], errors="coerce").fillna(0).iloc[0])
 
-    final_sekolah = _mland(_get_final("sekolah")) if False else _get_final("sekolah")
+    final_sekolah = _get_final("sekolah")
     final_umum    = _get_final("umum")
     final_khusus  = _get_final("khusus")
-    final_all     = (final_sekolah + final_umum + final_khusus) / 3.0
+    final_all     = (final_sekolah + final_umum + final_khusus) / 3.0  # sumber KPI final
 
+    # dasar = rata-rata 3 jenis (missing=0, tetap ÷3)
     def _get_dasar(j):
         if agg_jenis is None or agg_jenis.empty:
             return 0.0
@@ -1141,6 +1112,7 @@ def compute_dashboard_kpis(summary_jenis: pd.DataFrame, agg_total: pd.DataFrame,
     dasar_khusus  = _get_dasar("khusus")
     dasar_all     = (dasar_sekolah + dasar_umum + dasar_khusus) / 3.0
 
+    # cakupan
     if agg_total is not None and not agg_total.empty:
         n_sum = float(pd.to_numeric(agg_total.get("n_total", 0), errors="coerce").fillna(0).sum())
         t_ser = pd.to_numeric(agg_total.get("target_total_68", np.nan), errors="coerce")
@@ -1201,7 +1173,7 @@ def build_kpi_markdown(summary_jenis: pd.DataFrame, agg_total: pd.DataFrame, agg
 
 
 # ============================================================
-# 13) LLM + WORD (Word pakai Ringkasan Dashboard)
+# 13) LLM + WORD (Word pakai Ringkasan Dashboard, tanpa "Kewenangan: ...")
 # ============================================================
 
 _HF_CLIENT = None
@@ -1225,7 +1197,6 @@ def build_context(summary_jenis: pd.DataFrame, agg_total: pd.DataFrame, verif_to
     lines.append("Rumus penyesuaian: faktor = min(total_terkumpul / target_total_68, 1.0); Indeks_Final = Indeks_Dasar_Agregat × faktor.")
     lines.append("Jenis yang ditampilkan: sekolah, umum, khusus (SEMUA jenis menggunakan faktor wilayah).")
     lines.append("Catatan ringkasan: nilai 'keseluruhan' dihitung sebagai rata-rata 3 jenis (sekolah+umum+khusus) ÷ 3, missing=0.")
-    lines.append("Catatan konsistensi: Indeks_Dasar_Agregat wilayah juga dihitung sebagai rata-rata 3 jenis ÷3 (missing=0).")
 
     if summary_jenis is not None and not summary_jenis.empty:
         lines.append("\nRingkasan (jenis + keseluruhan):")
@@ -1297,14 +1268,16 @@ def generate_word_report(wilayah, summary_jenis, agg_total, agg_jenis, analysis_
     doc = Document()
     doc.add_heading(f"Laporan IPLM — {wilayah}", level=1)
 
+    # Ringkasan Dashboard (mengganti "Kewenangan: ...")
     doc.add_heading("Ringkasan Dashboard", level=2)
-    k = compute_dashboard_kpis(summary_jenis, agg_total, agg_jenis)
 
+    k = compute_dashboard_kpis(summary_jenis, agg_total, agg_jenis)
     doc.add_paragraph(f"Indeks IPLM FINAL (Disesuaikan): {k['final_all']:.2f} (rata-rata 3 jenis, tetap ÷3; missing=0)")
     doc.add_paragraph(f"Indeks Dasar (Tanpa Penyesuaian): {k['dasar_all']:.2f} (rata-rata 3 jenis, tetap ÷3; missing=0)")
     doc.add_paragraph(f"Cakupan Sampel (berdasarkan target 68%): {k['cakupan_pct']:.0f}% (min(total_terkumpul/target_68, 1.0))")
     doc.add_paragraph(f"Penyesuaian Nilai (rata-rata): {k['dampak']:.2f} poin (faktor penyesuaian mean: {k['faktor_mean']:.3f})")
 
+    # Tabel Ringkasan (Jenis + Keseluruhan)
     doc.add_paragraph("Ringkasan (Jenis + Keseluruhan):")
     show = summary_jenis.copy()
     preferred = [
@@ -1337,21 +1310,22 @@ def generate_word_report(wilayah, summary_jenis, agg_total, agg_jenis, analysis_
             else:
                 cells[i].text = str(v)
 
+    # Metodologi
     doc.add_heading("Metodologi", level=2)
     doc.add_paragraph(
         "Indeks dasar dihitung per entitas menggunakan transformasi Yeo-Johnson dan normalisasi MinMax nasional per indikator. "
-        "Nilai kemudian diagregasi per wilayah untuk memperoleh Indeks Dasar wilayah."
-    )
-    doc.add_paragraph(
-        "Indeks Dasar wilayah (keseluruhan) dihitung sebagai rata-rata 3 jenis perpustakaan (sekolah+umum+khusus) ÷ 3, "
-        "dengan missing dianggap 0, agar konsisten antar level analisis."
+        "Nilai kemudian diagregasi per wilayah (rata-rata) untuk memperoleh Indeks Dasar wilayah."
     )
     doc.add_paragraph(
         "Penyesuaian dilakukan berbasis kecukupan sampel minimum 68% pada level wilayah, "
         "dengan rumus faktor = min(total_terkumpul / target_total_68, 1.0). "
         "Indeks_Final_Wilayah = Indeks_Dasar_Agregat × faktor."
     )
+    doc.add_paragraph(
+        "Ringkasan 'keseluruhan' dihitung sebagai rata-rata 3 jenis (sekolah+umum+khusus) ÷ 3, dengan missing dianggap 0."
+    )
 
+    # Analisis LLM
     doc.add_heading("Analisis Naratif (LLM)", level=2)
     for p in (analysis_text or "").split("\n"):
         if p.strip():
@@ -1370,7 +1344,7 @@ def _empty_outputs(msg="⚠️ Data belum siap."):
     empty = pd.DataFrame()
     empty_fig = go.Figure()
     return (
-        "",
+        "",  # kpi_md
         empty, empty, empty, empty, empty,
         None, None, None, None, None,
         empty_fig, empty_fig, empty_fig,
@@ -1382,6 +1356,7 @@ def run_calc(prov_value, kab_value, kew_value, df_all, df_raw, pop_kab, pop_prov
         if df_all is None or df_all.empty or df_raw is None or df_raw.empty:
             return _empty_outputs("⚠️ Data belum ter-load. Pastikan file tersedia di repo/server.")
 
+        # FILTER ANALISIS (df_all)
         df = df_all.copy()
         if prov_value and prov_value != "(Semua)":
             df = df[df["PROV_DISP"] == prov_value]
@@ -1399,7 +1374,7 @@ def run_calc(prov_value, kab_value, kew_value, df_all, df_raw, pop_kab, pop_prov
         verif_total = build_verif_total(agg_total)
         detail_view = attach_final_to_detail(df, agg_total, meta, kew_value or "(Semua)")
 
-        # RAW download (df_raw)
+        # FILTER RAW DOWNLOAD (df_raw)
         raw = df_raw.copy()
         if prov_value and prov_value != "(Semua)":
             raw = raw[raw["PROV_DISP"] == prov_value]
@@ -1408,7 +1383,7 @@ def run_calc(prov_value, kab_value, kew_value, df_all, df_raw, pop_kab, pop_prov
         if kew_value and kew_value != "(Semua)":
             raw = raw[raw["KEW_NORM"] == kew_value]
 
-        # Bell curve per jenis (entitas)
+        # Bell curve per JENIS (per entitas)
         if detail_view is None or detail_view.empty:
             fig_sekolah = _make_bell_curve(pd.DataFrame(), "Indeks_Dasar_0_100", "Bell Curve — Jenis: Sekolah", min_points=2)
             fig_umum    = _make_bell_curve(pd.DataFrame(), "Indeks_Dasar_0_100", "Bell Curve — Jenis: Umum", min_points=2)
@@ -1426,6 +1401,7 @@ def run_calc(prov_value, kab_value, kew_value, df_all, df_raw, pop_kab, pop_prov
             fig_umum    = _fig_jenis_ent("umum",    "Bell Curve — Jenis: Umum (Indeks per Entitas)")
             fig_khusus  = _fig_jenis_ent("khusus",  "Bell Curve — Jenis: Khusus (Indeks per Entitas)")
 
+        # KPI markdown (FINAL sumber Ringkasan)
         kpi_md = build_kpi_markdown(summary_jenis, agg_total, agg_jenis)
 
         tmpdir = tempfile.mkdtemp()
@@ -1435,30 +1411,31 @@ def run_calc(prov_value, kab_value, kew_value, df_all, df_raw, pop_kab, pop_prov
 
         p_summary = str(Path(tmpdir) / f"IPLM_RingkasanJenisKeseluruhan_{prov_slug}_{kab_slug}_{kew_slug}.xlsx")
         p_total   = str(Path(tmpdir) / f"IPLM_AgregatWilayah_Keseluruhan_{prov_slug}_{kab_slug}_{kew_slug}.xlsx")
-        p_raw     = str(Path(tmpdir) / f"IPLM_RAW_DATA_{prov_slug}_{kab_slug}_{kew_slug}.xlsx")
+        p_jenis   = str(Path(tmpdir) / f"IPLM_RAW_DATA_{prov_slug}_{kab_slug}_{kew_slug}.xlsx")
         p_detail  = str(Path(tmpdir) / f"IPLM_DetailEntitas_FinalMenempelWilayah_{prov_slug}_{kab_slug}_{kew_slug}.xlsx")
         p_verif   = str(Path(tmpdir) / f"IPLM_KecukupanSampel68_{prov_slug}_{kab_slug}_{kew_slug}.xlsx")
 
         summary_jenis.to_excel(p_summary, index=False)
         agg_total.to_excel(p_total, index=False)
-        raw.to_excel(p_raw, index=False)
+        raw.to_excel(p_jenis, index=False)
         detail_view.to_excel(p_detail, index=False)
         verif_total.to_excel(p_verif, index=False)
 
         wilayah_txt = kab_value if (kab_value and kab_value != "(Semua)") else (prov_value if (prov_value and prov_value != "(Semua)") else "Nasional/All")
         analysis_text = generate_llm_analysis(summary_jenis, agg_total, verif_total, wilayah_txt, kew_value or "(Semua)")
+
+        # Word report: tanpa "Kewenangan: ..."
         word_path = generate_word_report(wilayah_txt, summary_jenis, agg_total, agg_jenis, analysis_text)
 
         msg = (
             f"✅ Selesai: raw={len(raw)} | entitas={len(detail_view)} | wilayah(keseluruhan)={len(agg_total)} | "
-            f"jenis(agregat)={len(agg_jenis)} | ringkasan: 3 jenis selalu tampil & keseluruhan ÷3 | "
-            f"CONSISTENCY FIX: indeks wilayah keseluruhan juga ÷3"
+            f"jenis(agregat)={len(agg_jenis)} | ringkasan: 3 jenis selalu tampil & keseluruhan ÷3"
         )
 
         return (
             kpi_md,
             summary_jenis, agg_total, agg_jenis, detail_view, verif_total,
-            p_summary, p_total, p_raw, p_detail, word_path,
+            p_summary, p_total, p_jenis, p_detail, word_path,
             fig_umum, fig_sekolah, fig_khusus,
             msg, analysis_text
         )
@@ -1468,7 +1445,7 @@ def run_calc(prov_value, kab_value, kew_value, df_all, df_raw, pop_kab, pop_prov
 
 
 # ============================================================
-# 15) UI (NO UPLOAD) — LANJUTAN (tidak terputus)
+# 15) UI (NO UPLOAD)
 # ============================================================
 
 def ui_load(force=False):
@@ -1485,172 +1462,118 @@ def ui_load(force=False):
     prov_vals = [v for v in prov_vals if v and v.strip()]
     prov_choices = ["(Semua)"] + sorted(set(prov_vals))
 
-    kew_vals = df_all["KEW_NORM"].dropna().astype(str).tolist()
-    kew_vals = [v for v in kew_vals if v and v.strip()]
-    kew_choices = ["(Semua)"] + sorted(set(kew_vals))
-
-    kab_vals = df_all["KAB_DISP"].dropna().astype(str).tolist()
-    kab_vals = [v for v in kab_vals if v and v.strip()]
-    kab_choices = ["(Semua)"] + sorted(set(kab_vals))
+    kab_choices  = ["(Semua)"] + sorted([x for x in df_all["KAB_DISP"].dropna().unique().tolist() if x])
+    kew_choices  = ["(Semua)"] + sorted([x for x in df_all["KEW_NORM"].dropna().unique().tolist() if x])
+    default_kew  = "PROVINSI" if "PROVINSI" in kew_choices else ("KAB/KOTA" if "KAB/KOTA" in kew_choices else "(Semua)")
 
     return (
         df_all, df_raw, pop_kab, pop_prov, pop_khusus, meta, info,
         gr.update(choices=prov_choices, value="(Semua)"),
         gr.update(choices=kab_choices, value="(Semua)"),
-        gr.update(choices=kew_choices, value="(Semua)"),
+        gr.update(choices=kew_choices, value=default_kew),
     )
 
-def ui_update_kab_choices(prov_value, df_all):
+def on_prov_change(prov_value):
+    df_all, _, _, _, _, _, _ = load_default_files(force=False)
     if df_all is None or df_all.empty:
         return gr.update(choices=["(Semua)"], value="(Semua)")
-    if prov_value and prov_value != "(Semua)":
-        sub = df_all[df_all["PROV_DISP"] == prov_value]
+    if prov_value is None or prov_value == "(Semua)":
+        vals = df_all["KAB_DISP"].dropna().unique().tolist()
     else:
-        sub = df_all
-    kabs = sub["KAB_DISP"].dropna().astype(str).tolist()
-    kabs = [k for k in kabs if k and k.strip()]
-    kab_choices = ["(Semua)"] + sorted(set(kabs))
-    return gr.update(choices=kab_choices, value="(Semua)")
-
-def ui_run(prov_value, kab_value, kew_value, df_all, df_raw, pop_kab, pop_prov, pop_khusus, meta):
-    return run_calc(prov_value, kab_value, kew_value, df_all, df_raw, pop_kab, pop_prov, pop_khusus, meta)
-
+        vals = df_all.loc[df_all["PROV_DISP"] == prov_value, "KAB_DISP"].dropna().unique().tolist()
+    vals = sorted([v for v in vals if v])
+    return gr.update(choices=["(Semua)"] + vals, value="(Semua)")
+
+
+with gr.Blocks() as demo:
+    gr.Markdown(f"""
+# IPLM 2025 — Final (Penyesuaian Berbasis Kecukupan Sampel 68%)
+**Mode NO UPLOAD (cache aktif).** File dibaca dari repo/server:
+- `DATA_FILE`   = **{DATA_FILE}**
+- `POP_KAB`     = **{POP_KAB}**
+- `POP_PROV`    = **{POP_PROV}**
+- `POP_KHUSUS`  = **{POP_KHUSUS}**
+
+**Update FIX (konsistensi dashboard):**
+- Ringkasan selalu tampil **sekolah, umum, khusus, keseluruhan** (walau 0)
+- **Keseluruhan = rata-rata 3 jenis ÷3 (missing=0)**
+- KPI FINAL dashboard sumber dari Ringkasan (bukan mean baris yang ada)
+- Download Data Mentah = RAW hasil filter
+- Laporan Word: tanpa baris "Kewenangan: ...", diganti "Ringkasan Dashboard"
+""")
+
+    state_df = gr.State(None)
+    state_raw = gr.State(None)
+    state_pop_kab = gr.State(None)
+    state_pop_prov = gr.State(None)
+    state_pop_khusus = gr.State(None)
+    state_meta = gr.State({})
+
+    info_box = gr.Markdown()
 
-# ============================================================
-# 16) BUILD APP
-# ============================================================
+    with gr.Row():
+        dd_prov = gr.Dropdown(label="Provinsi", choices=["(Semua)"], value="(Semua)")
+        dd_kab  = gr.Dropdown(label="Kab/Kota", choices=["(Semua)"], value="(Semua)")
+        dd_kew  = gr.Dropdown(label="Kewenangan", choices=["(Semua)"], value="(Semua)")
 
-with gr.Blocks(title="IPLM 2025 — Dashboard (NO UPLOAD)") as demo:
-
-    # ---------- STATE ----------
-    st_df_all     = gr.State(None)
-    st_df_raw     = gr.State(None)
-    st_pop_kab    = gr.State(None)
-    st_pop_prov   = gr.State(None)
-    st_pop_khusus = gr.State(None)
-    st_meta       = gr.State({})
-    st_info       = gr.State("")
-
-    # ---------- HEADER ----------
-    gr.Markdown(
-        """
-# IPLM 2025 — Dashboard (NO UPLOAD)
-**Konsistensi hitung keseluruhan wilayah (Kab/Kota/Provinsi) vs individu dijamin dengan:**
-- Indeks dasar keseluruhan wilayah = (dasar_sekolah + dasar_umum + dasar_khusus) ÷ 3 (missing=0, tetap ÷3)
-- Faktor penyesuaian = min(total_terkumpul / target_total_68, 1.0)
-- Indeks final wilayah = indeks dasar keseluruhan × faktor penyesuaian
-        """.strip()
-    )
+    dd_prov.change(fn=on_prov_change, inputs=[dd_prov], outputs=dd_kab)
 
-    info_box = gr.HTML(value="")
+    run_btn = gr.Button("Jalankan Perhitungan")
+    msg_out = gr.Markdown()
 
-    with gr.Row():
-        prov_dd = gr.Dropdown(label="Provinsi", choices=["(Semua)"], value="(Semua)", interactive=True)
-        kab_dd  = gr.Dropdown(label="Kab/Kota", choices=["(Semua)"], value="(Semua)", interactive=True)
-        kew_dd  = gr.Dropdown(label="Kewenangan", choices=["(Semua)"], value="(Semua)", interactive=True)
+    kpi_out = gr.Markdown()
 
-    with gr.Row():
-        btn_reload = gr.Button("Reload File (Cache)", variant="secondary")
-        btn_run    = gr.Button("Run", variant="primary")
-
-    # ---------- KPI ----------
-    kpi_html = gr.HTML(value="")
-
-    # ---------- TABLES ----------
-    with gr.Tabs():
-        with gr.Tab("Ringkasan (Jenis + Keseluruhan)"):
-            tbl_summary = gr.Dataframe(
-                headers=["Jenis","Jumlah_Wilayah","Total_Perpus","Rata2_dim_kepatuhan","Rata2_dim_kinerja","Indeks_Final_Disesuaikan_0_100"],
-                interactive=False
-            )
+    gr.Markdown("## Ringkasan (Jenis + Keseluruhan)")
+    out_summary = gr.DataFrame(interactive=False)
 
-        with gr.Tab("Agregat Wilayah (Keseluruhan)"):
-            tbl_agg_total = gr.Dataframe(interactive=False)
+    gr.Markdown("## Agregat Wilayah (Keseluruhan)")
+    out_agg_total = gr.DataFrame(interactive=False)
 
-        with gr.Tab("Agregat Wilayah × Jenis"):
-            tbl_agg_jenis = gr.Dataframe(interactive=False)
+    gr.Markdown("## Agregat Wilayah × Jenis (Sekolah, Umum, Khusus)")
+    out_agg_jenis = gr.DataFrame(interactive=False)
 
-        with gr.Tab("Detail Entitas (Final menempel Wilayah)"):
-            tbl_detail = gr.Dataframe(interactive=False)
+    gr.Markdown("## Detail Entitas (Final menempel dari wilayah)")
+    out_detail = gr.DataFrame(interactive=False)
 
-        with gr.Tab("Verifikasi 68% (Total Wilayah)"):
-            tbl_verif = gr.Dataframe(interactive=False)
+    gr.Markdown("## Kecukupan Sampel 68% (tanpa angka koma)")
+    out_verif = gr.DataFrame(interactive=False)
 
-    # ---------- DOWNLOADS ----------
-    with gr.Row():
-        dl_summary = gr.File(label="Download Ringkasan (xlsx)")
-        dl_total   = gr.File(label="Download Agregat Keseluruhan (xlsx)")
-        dl_raw     = gr.File(label="Download RAW Data (xlsx)")
-        dl_detail  = gr.File(label="Download Detail Entitas (xlsx)")
-        dl_word    = gr.File(label="Download Word Report (docx)")
-
-    # ---------- CHARTS ----------
-    with gr.Tabs():
-        with gr.Tab("Bell Curve — Umum"):
-            fig_umum = gr.Plot()
-        with gr.Tab("Bell Curve — Sekolah"):
-            fig_sekolah = gr.Plot()
-        with gr.Tab("Bell Curve — Khusus"):
-            fig_khusus = gr.Plot()
-
-    # ---------- LLM OUTPUT ----------
-    with gr.Accordion("Analisis Otomatis (LLM)", open=False):
-        status_text = gr.Markdown("")
-        llm_text = gr.Markdown("")
-
-    # ============================================================
-    # EVENTS
-    # ============================================================
-
-    def _on_load():
-        df_all, df_raw, pop_kab, pop_prov, pop_khusus, meta, info, prov_upd, kab_upd, kew_upd = ui_load(force=False)
-        return (
-            df_all, df_raw, pop_kab, pop_prov, pop_khusus, meta, info,
-            info, prov_upd, kab_upd, kew_upd
-        )
+    gr.Markdown("## Bell Curve — per Jenis Perpustakaan (Indeks per Entitas)")
+    gr.Markdown("### Perpustakaan Umum")
+    bell_umum = gr.Plot(scale=1)
 
-    demo.load(
-        _on_load,
-        inputs=[],
-        outputs=[st_df_all, st_df_raw, st_pop_kab, st_pop_prov, st_pop_khusus, st_meta, st_info,
-                 info_box, prov_dd, kab_dd, kew_dd]
-    )
+    gr.Markdown("### Perpustakaan Sekolah")
+    bell_sekolah = gr.Plot(scale=1)
 
-    btn_reload.click(
-        lambda: (
-            *ui_load(force=True)[:7],
-            ui_load(force=True)[6],  # info
-            ui_load(force=True)[7],  # prov upd
-            ui_load(force=True)[8],  # kab upd
-            ui_load(force=True)[9],  # kew upd
-        ),
-        inputs=[],
-        outputs=[st_df_all, st_df_raw, st_pop_kab, st_pop_prov, st_pop_khusus, st_meta, st_info,
-                 info_box, prov_dd, kab_dd, kew_dd]
-    )
+    gr.Markdown("### Perpustakaan Khusus")
+    bell_khusus = gr.Plot(scale=1)
 
-    prov_dd.change(
-        ui_update_kab_choices,
-        inputs=[prov_dd, st_df_all],
-        outputs=[kab_dd]
-    )
+    gr.Markdown("## Analisis Otomatis (LLM)")
+    analysis_out = gr.Markdown()
 
-    btn_run.click(
-        ui_run,
-        inputs=[prov_dd, kab_dd, kew_dd, st_df_all, st_df_raw, st_pop_kab, st_pop_prov, st_pop_khusus, st_meta],
+    with gr.Row():
+        dl_summary = gr.DownloadButton(label="Download Ringkasan (.xlsx)")
+        dl_total   = gr.DownloadButton(label="Download Agregat Wilayah (.xlsx)")
+        dl_jenis   = gr.DownloadButton(label="Download Data Mentah (.xlsx)")
+        dl_detail  = gr.DownloadButton(label="Download Detail Entitas (.xlsx)")
+        dl_word    = gr.DownloadButton(label="Download Laporan Word (.docx)")
+
+    run_btn.click(
+        fn=run_calc,
+        inputs=[dd_prov, dd_kab, dd_kew, state_df, state_raw, state_pop_kab, state_pop_prov, state_pop_khusus, state_meta],
         outputs=[
-            kpi_html,
-            tbl_summary, tbl_agg_total, tbl_agg_jenis, tbl_detail, tbl_verif,
-            dl_summary, dl_total, dl_raw, dl_detail, dl_word,
-            fig_umum, fig_sekolah, fig_khusus,
-            status_text, llm_text
+            kpi_out,
+            out_summary, out_agg_total, out_agg_jenis, out_detail, out_verif,
+            dl_summary, dl_total, dl_jenis, dl_detail, dl_word,
+            bell_umum, bell_sekolah, bell_khusus,
+            msg_out, analysis_out
         ]
     )
 
+    demo.load(
+        fn=lambda: ui_load(force=False),
+        inputs=[],
+        outputs=[state_df, state_raw, state_pop_kab, state_pop_prov, state_pop_khusus, state_meta, info_box, dd_prov, dd_kab, dd_kew]
+    )
 
-# ============================================================
-# 17) LAUNCH
-# ============================================================
-
-if __name__ == "__main__":
-    demo.launch(share=True)
+demo.launch()