fix: performance issue in interpretability notebooks (#1238)

microsoft · Nov 3, 2021 · 5733b85 · 5733b85
1 parent 81f5f80
commit 5733b85
Show file tree

Hide file tree

Showing 2 changed files with 2 additions and 2 deletions.
diff --git a/notebooks/Interpretability - Explanation Dashboard.ipynb b/notebooks/Interpretability - Explanation Dashboard.ipynb
@@ -169,7 +169,7 @@
     "    model=model,\n",
     "    targetCol=\"probability\",\n",
     "    targetClasses=[1],\n",
-    "    backgroundData=training.orderBy(rand()).limit(100).cache(),\n",
+    "    backgroundData=broadcast(training.orderBy(rand()).limit(100).cache()),\n",
     ")\n",
     "\n",
     "shap_df = shap.transform(explain_instances)\n"

diff --git a/notebooks/Interpretability - Tabular SHAP explainer.ipynb b/notebooks/Interpretability - Tabular SHAP explainer.ipynb
@@ -165,7 +165,7 @@
     "    model=model,\n",
     "    targetCol=\"probability\",\n",
     "    targetClasses=[1],\n",
-    "    backgroundData=training.orderBy(rand()).limit(100).cache(),\n",
+    "    backgroundData=broadcast(training.orderBy(rand()).limit(100).cache()),\n",
     ")\n",
     "\n",
     "shap_df = shap.transform(explain_instances)\n"