upgrade code for retrofit walls and roofs calls

src/features/build_features.py

@@ -66,7 +66,7 @@ energy_system_columns = [
 
 envelop_columns = [
     "FLOOR_U_VALUE","FLOOR_INSULATION_TYPE","FLOOR_BOUNDARY_TYPE",
-    "WALL_U_VALUE","WALL_TYPE","WALL_INSULATION",
+    "WALL_U_VALUE","WALL_TYPE","WALL_INSULATION_MODEL",
     "ROOF_U_VALUE","ROOF_CLASS","ROOF_INSULATION_TYPE",
     "glazing_area_m2","glazing_type"
 ]
@@ -83,7 +83,7 @@ cat_cols = [
     "MAIN_HEATING_SYSTEM","SECONDARY_HEATING_SYSTEM",
     "MAIN_FUEL_TYPE","DHW_SUPPLY_SYSTEM","VENTILATION_SYSTEM",
     "FLOOR_INSULATION_TYPE","FLOOR_BOUNDARY_TYPE",
-    "WALL_TYPE","WALL_INSULATION",
+    "WALL_TYPE","WALL_INSULATION_MODEL",
     "ROOF_CLASS","ROOF_INSULATION_TYPE",
     "glazing_type",
     "PROPERTY_TYPE","BUILT_FORM","sap_band_letter"

src/features/roofs.py

@@ -313,6 +313,11 @@ def extract_roof_insulation(row):
     return "unknown"
 
 
+S9_MM = np.array([0, 12, 25, 50, 75, 100, 150, 200, 250, 270, 300, 350, 400])
+S9_U  = np.array([2.3, 1.5, 1.0, 0.68, 0.50, 0.40, 0.30, 0.21, 0.17, 0.16, 0.14, 0.12, 0.11])
+
+S9_LOOKUP = dict(zip(S9_MM, S9_U))
+
 
 def build_roof_lookup(roof_desc: pd.Series) -> pd.DataFrame:
     """
@@ -356,6 +361,34 @@ def build_roof_lookup(roof_desc: pd.Series) -> pd.DataFrame:
     mm = s.str.extract(r"(\d+)\s*\+?\s*mm", expand=False)
     out["ROOF_MM_RAW"] = pd.to_numeric(mm, errors="coerce")
 
+    # ---------------------------
+    # APPLY RETROFIT TO MEASURED U-VALUES
+    # ---------------------------
+
+    mask_measured_upgrade = (
+            out["ROOF_MEASURED_U"].notna() &
+            out["ROOF_MM_RAW"].notna()
+    )
+
+    if mask_measured_upgrade.any():
+        u_meas = out.loc[mask_measured_upgrade, "ROOF_MEASURED_U"].values
+        mm_add = out.loc[mask_measured_upgrade, "ROOF_MM_RAW"].astype(int).values
+
+        # inverse S9 (nearest)
+        diff = np.abs(u_meas[:, None] - S9_U[None, :])
+        base_mm = S9_MM[diff.argmin(axis=1)]
+
+        # add retrofit + clip
+        new_mm = np.minimum(base_mm + mm_add, 400)
+
+        # forward S9 lookup
+        out.loc[mask_measured_upgrade, "ROOF_MEASURED_U"] = S9_U[
+            np.searchsorted(S9_MM, new_mm)
+        ]
+
+    # zero out insulation thickness for measured U-value rows
+    out.loc[out["ROOF_CLASS"] == "measured", "ROOF_MM_RAW"] = pd.NA
+
     # ---------------------------
     # NORMALISE TO SAP S9 MM
     # ---------------------------

src/features/walls.py

@@ -6,31 +6,61 @@ import numpy as np
 def extract_wall_u_from_text(text: str | float | None) -> float | None:
     """
     Extract numeric U-value from WALLS_DESCRIPTION when it contains
-    'Average thermal transmittance ...', which corresponds to the
-    average external wall U-value in SAP/RdSAP EPCs.
+    'Average thermal transmittance ...'.
 
-    If no numeric value is found or it looks invalid (< 0.05),
-    return None and let Stage 2 handle it.
+    Supports optional insulation thickness suffix:
+        '..., 0 mm'
+        '..., 50 mm'
+        '..., 100 mm'
+        etc.
+
+    If insulation is present, applies R-addition.
     """
     if pd.isna(text):
         return None
 
     s = str(text).lower()
+
     if "average thermal transmittance" not in s:
         return None
 
-    # Find first number in the string (handles '0.30', '1.4', etc.)
+    # ------------------------------------------------------------
+    # 1. Extract baseline U-value
+    # ------------------------------------------------------------
     nums = re.findall(r"([0-9]*\.?[0-9]+)", s)
     if not nums:
         return None
 
-    u = float(nums[0])
+    u_base = float(nums[0])
 
-    # EPC sometimes has '0.00' for missing; treat as unknown
-    if u < 0.05:
+    # EPC sometimes has '0.00' for missing
+    if u_base < 0.05:
         return None
 
-    return u
+    # ------------------------------------------------------------
+    # 2. Extract insulation thickness (mm), default = 0 mm
+    # ------------------------------------------------------------
+    mm_match = re.search(r"(\d+)\s*mm", s)
+    mm = int(mm_match.group(1)) if mm_match else 0
+
+    # ------------------------------------------------------------
+    # 3. Apply R-addition if insulation present
+    # ------------------------------------------------------------
+    R_INS_MAP = {
+        0:   0.0,
+        50:  1.4,
+        100: 2.8,
+        150: 4.2,
+        200: 5.6,
+    }
+
+    R_ins = R_INS_MAP.get(mm, 0.0)
+
+    if R_ins > 0:
+        R_old = 1.0 / u_base
+        return 1.0 / (R_old + R_ins)
+
+    return u_base
 
 
 
@@ -368,9 +398,24 @@ def wall_feature_engineering(
     # ------------------------------------------------------------
     df["WALL_U_VALUE"] = df["WALL_U_MEASURED"].combine_first(df["WALL_U_TABLE"])
 
+    # ------------------------------------------------------------
+    # 4.5 Vectorised insulation collapse for ML model
+    # ------------------------------------------------------------
+
+    # Start with default = insulated
+    df["WALL_INSULATION_MODEL"] = "insulated"
+
+    # as built → as built
+    mask_as_built = df["WALL_INSULATION"].isin(["as built"])
+    df.loc[mask_as_built, "WALL_INSULATION_MODEL"] = "as built"
+
+    # unknown / NaN → unknown
+    mask_unknown = df["WALL_INSULATION"].isna() | df["WALL_INSULATION"].isin(["unknown"])
+    df.loc[mask_unknown, "WALL_INSULATION_MODEL"] = "unknown"
+
     # ------------------------------------------------------------
     # 5. Optional clean-up
     # ------------------------------------------------------------
-    df.drop(columns=["WALL_U_TABLE"], inplace=True, errors="ignore")
+    df.drop(columns=["WALL_U_TABLE","WALL_INSULATION"], inplace=True, errors="ignore")
 
     return df