Update app.py

app.py

@@ -67,27 +67,32 @@ def call_ai_model(usage, idle, freq, cost, last):
         cost = float(cost) if cost is not None and not np.isnan(cost) else 0.0
 
         total = usage + idle
-        ratio = usage / total if total > 0 else 0
-        if ratio < 0.3:
+        ratio = usage / total if total > 0 else 0.0
+        utilization_percent = ratio * 100
+
+        # AI Suggestion logic based on utilization percent
+        if utilization_percent < 30:
             sug = "Pause Rent"
-        elif ratio < 0.6:
+        elif utilization_percent < 60:
             sug = "Move"
-        elif ratio < 0.8:
+        elif utilization_percent < 80:
             sug = "Repair"
         else:
             sug = "Replace"
 
-        # Updated confidence logic:
+        # Confidence calculation
+        base_conf = min(ratio + 0.1, 1.0)
         if idle > usage:
-            conf = max(0.1, (usage / (idle + usage)) * 0.5)  # Lower confidence if idle dominates
-        else:
-            conf = min(1.0, (usage / (idle + usage)) + 0.5)  # Higher confidence if usage dominates
-
-        score = ratio * 100
+            base_conf *= 0.6
+        freq_factor = min(freq / 10.0, 1.0)
+        base_conf *= (0.7 + 0.3 * freq_factor)
+        confidence = max(0.1, min(base_conf, 1.0))
 
-        if np.isnan(conf) or np.isnan(score):
+        if np.isnan(confidence) or np.isnan(utilization_percent):
             raise ValueError("Computed values resulted in NaN. Please check input data.")
-        return sug, conf, score
+
+        return sug, confidence, utilization_percent
+
     except Exception as e:
         logger.error(f"Error in call_ai_model: {e}")
         raise ValueError(f"AI model computation failed: {str(e)}")
@@ -414,4 +419,4 @@ with gr.Blocks() as app:
 #result-box { border: 3px solid #d3d3d3 !important; border-radius: 10px; padding: 10px; background: #f9f9f9; white-space: pre-line; }
 """
 if __name__ == "__main__":
-    app.launch()
+    app.launch()