Update app.py

app.py

@@ -23,7 +23,7 @@ from sklearn.metrics import mean_squared_error  # MAE removed
 APP_NAME = "ST_TOC"
 TAGLINE  = "Total Organic Carbon Estimation Using AI"
 
-FEATURES = ["AHT90 (Average Hydrocarbon Tool 90° Phase)", "DT (Delta-T Sonic Travel Time)", "GR (Gamma Ray)", "K (Potassium)", "RHOB (Bulk Density)", "TNPH (Thermal Neutron Porosity)", "Th (Thorium)", "Ur (Uranium)"]
+FEATURES = ["AHT90", "DT", "GR", "K", "RHOB", "TNPH", "Th", "Ur"]
 TARGET    = "TOC"
 PRED_COL  = "TOC_Pred"
 
@@ -157,35 +157,50 @@ def read_book_bytes(b: bytes):
 
 # ---- Canonical feature aliasing (accept legacy headers) ----
 def _build_alias_map(canonical_features: list[str], target_name: str) -> dict:
+    # Helper: prefer a canonical name from FEATURES if present
     def pick(expected, variants):
         for v in variants:
-            if v in expected: return v
-        return variants[0]
-
-    # TOC logs/features
-    can_GR   = pick(canonical_features, ["GR", "Gamma Ray", "GR (API)"])
-    can_RHOB = pick(canonical_features, ["RHOB", "Bulk Density", "RHOB (g/cc)"])
-    can_DT   = pick(canonical_features, ["DT", "AC", "DT (us/ft)"])
-    can_TNPH = pick(canonical_features, ["TNPH", "NPHI", "TNPH (%)"])
-    can_K    = pick(canonical_features, ["K", "Potassium", "K (%)"])
-    can_Th   = pick(canonical_features, ["Th", "Thorium", "Th (ppm)"])
-    can_Ur   = pick(canonical_features, ["Ur", "U", "U (ppm)"])
-    can_AHT  = pick(canonical_features, ["AHT90", "AHT_90", "AHT-90"])
+            if v in expected:
+                return v
+        return variants[0]  # fall back
+
+    # Canonical abbreviations (what we WANT in the DataFrame)
+    can_AHT  = pick(canonical_features, ["AHT90"])
+    can_DT   = pick(canonical_features, ["DT"])
+    can_GR   = pick(canonical_features, ["GR"])
+    can_K    = pick(canonical_features, ["K"])
+    can_RHOB = pick(canonical_features, ["RHOB"])
+    can_TNPH = pick(canonical_features, ["TNPH"])
+    can_Th   = pick(canonical_features, ["Th"])
+    can_Ur   = pick(canonical_features, ["Ur"])
 
     alias = {
+        # Abbrev ↔ variants (map everything to the abbrev)
+        "AHT90": can_AHT, "AHT_90": can_AHT, "AHT-90": can_AHT,
+        "AHT90 (Average Hydrocarbon Tool 90° Phase)": can_AHT,
+
+        "DT": can_DT, "AC": can_DT, "DT (us/ft)": can_DT,
+        "DT (Delta-T Sonic Travel Time)": can_DT,
+
         "GR": can_GR, "Gamma Ray": can_GR, "GR (API)": can_GR,
+        "GR (Gamma Ray)": can_GR,
+
+        "K": can_K, "Potassium": can_K, "K (%)": can_K,
+
         "RHOB": can_RHOB, "Bulk Density": can_RHOB, "RHOB (g/cc)": can_RHOB,
-        "DT": can_DT, "AC": can_DT, "DT (us/ft)": can_DT,
+        "RHOB (Bulk Density)": can_RHOB,
+
         "TNPH": can_TNPH, "NPHI": can_TNPH, "TNPH (%)": can_TNPH,
-        "K": can_K, "Potassium": can_K, "K (%)": can_K,
+        "TNPH (Thermal Neutron Porosity)": can_TNPH,
+
         "Th": can_Th, "Thorium": can_Th, "Th (ppm)": can_Th,
+
         "Ur": can_Ur, "U": can_Ur, "U (ppm)": can_Ur,
-        "AHT90": can_AHT, "AHT_90": can_AHT, "AHT-90": can_AHT,
 
-        # Optional depth for tracks
-        "Depth (ft)": "Depth (ft)", "DEPTH": "Depth (ft)", "MD (ft)": "Depth (ft)",
+        # Optional depth aliases (if you add later)
+        "Depth": "Depth", "Depth (ft)": "Depth", "DEPTH": "Depth", "MD (ft)": "Depth",
 
-        # Targets
+        # Target aliases → canonical TARGET
         "TOC": target_name, "TOC (%)": target_name, "Total Organic Carbon": target_name,
     }
     return alias
@@ -576,7 +591,7 @@ except Exception as e:
     st.error(f"Failed to load model: {e}")
     st.stop()
 
-# Prefer UCS-specific meta
+# Prefer toc-specific meta
 meta = {}
 meta_candidates = [MODELS_DIR / "toc_meta.json", MODELS_DIR / "meta.json"]
 meta_path = next((p for p in meta_candidates if p.exists()), None)