[Fix] Export IUCLID Properties, Show Model Params and Statistics, Adjust Batch Prediction Settings (#434)

Co-authored-by: Tim Lorsbach <tim@lorsba.ch>
Reviewed-on: enviPath/enviPy#434

epdb/models.py

@@ -2162,7 +2162,7 @@ class Pathway(EnviPathModel, AliasMixin, ScenarioMixin, AdditionalInformationMix
 
             row += [cs.smiles, cs.get_name(), n.depth]
 
-            edges = self.edges.filter(end_nodes__in=[n])
+            edges = self.edges.filter(end_nodes=n)
             if len(edges):
                 for e in edges:
                     _row = row.copy()
@@ -2847,6 +2847,58 @@ class PackageBasedModel(EPModel):
 
         return res
 
+    def parameters(self):
+        params = {
+            "Model Evaluation Threshold": f"{self.threshold:.2f}",
+            "Multi Gen Evaluation": "Yes" if self.multigen_eval else "No",
+        }
+
+        if self.app_domain:
+            params["Applicability Domain Num Neighbors"] = f"{self.app_domain.num_neighbours:.2f}"
+            params["Applicability Domain Reliability Threshold"] = (
+                f"{self.app_domain.reliability_threshold:.2f}"
+            )
+            params["Applicability Domain Local Compatibility Threshold"] = (
+                f"{self.app_domain.local_compatibilty_threshold:.2f}"
+            )
+
+        return params
+
+    def statistics(self):
+        from sklearn.metrics import auc
+
+        recall = list(self.eval_results["average_recall_per_threshold"].values())
+        precision = list(self.eval_results["average_precision_per_threshold"].values())
+        mg_recall = list(
+            self.eval_results.get("multigen_average_recall_per_threshold", {}).values()
+        )
+        mg_precision = list(
+            self.eval_results.get("multigen_average_precision_per_threshold", {}).values()
+        )
+
+        return {
+            "accuracy": [
+                self.eval_results["average_accuracy"],
+                self.eval_results.get("multigen_average_accuracy"),
+            ],
+            "precision": [
+                self.eval_results["average_precision_per_threshold"][f"{self.threshold:.2f}"],
+                self.eval_results.get("multigen_average_precision_per_threshold", {}).get(
+                    f"{self.threshold:.2f}"
+                ),
+            ],
+            "recall": [
+                self.eval_results["average_recall_per_threshold"][f"{self.threshold:.2f}"],
+                self.eval_results.get("multigen_average_recall_per_threshold", {}).get(
+                    f"{self.threshold:.2f}"
+                ),
+            ],
+            "Area under PR Curve": [
+                auc(recall, precision),
+                auc(mg_recall, mg_precision) if self.multigen_eval else None,
+            ],
+        }
+
     @cached_property
     def applicable_rules(self) -> List["Rule"]:
         """
@@ -3027,8 +3079,6 @@ class PackageBasedModel(EPModel):
                 thresholds.append(np.float64(threshold))
                 thresholds.sort()
 
-            logger.info(f"Thresholds: {thresholds}")
-
             precision = {f"{t:.2f}": [] for t in thresholds}
             recall = {f"{t:.2f}": [] for t in thresholds}
 
@@ -3077,14 +3127,17 @@ class PackageBasedModel(EPModel):
             for t in thresholds:
                 for true, pred in zip(pathways, pred_pathways):
                     acc, pre, rec = multigen_eval(true, pred, t)
-                    if abs(t - threshold) < 0.01:
-                        mg_acc = acc
+
+                    if f"{t:.2f}" == f"{threshold:.2f}":
+                        mg_acc += acc
+
                     precision[f"{t:.2f}"].append(pre)
                     recall[f"{t:.2f}"].append(rec)
 
+            avg_mg_acc = mg_acc / len(root_compounds)
             precision = {k: sum(v) / len(v) if len(v) > 0 else 0 for k, v in precision.items()}
             recall = {k: sum(v) / len(v) if len(v) > 0 else 0 for k, v in recall.items()}
-            return mg_acc, precision, recall
+            return avg_mg_acc, precision, recall
 
         # If there are eval packages perform single generation evaluation on them instead of random splits
         if self.eval_packages.count() > 0:
@@ -4216,6 +4269,9 @@ class ClassifierPluginModel(PackageBasedModel):
         instance = impl(conf)
         return instance
 
+    def parameters(self):
+        return self.instance().parameters()
+
     def build_dataset(self):
         """
         Required by general model contract but actual implementation resides in plugin.
@@ -4436,6 +4492,9 @@ class PropertyPluginModel(PackageBasedModel):
         instance = impl()
         return instance
 
+    def parameters(self):
+        return self.instance().parameters()
+
     def build_dataset(self):
         """
         Required by general model contract but actual implementation resides in plugin.

epdb/tasks.py

@@ -477,8 +477,7 @@ def batch_predict(
             limit=None,
             setting_overrides={
                 "max_nodes": num_tps,
-                "max_depth": num_tps,
-                "model_threshold": 0.001,
+                "model_threshold": 0.0,
             },
         )