Refactor for dowhy_confounder_example

py-why · Jul 2, 2024 · 6be4c37 · 6be4c37
1 parent 7ebe273
commit 6be4c37
Show file tree

Hide file tree

Showing 2 changed files with 79 additions and 9 deletions.
diff --git a/docs/source/example_notebooks/dowhy_confounder_example.ipynb b/docs/source/example_notebooks/dowhy_confounder_example.ipynb
@@ -123,14 +123,46 @@
     "Identify the causal effect using properties of the causal graph."
    ]
   },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "df"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": null,
    "metadata": {},
    "outputs": [],
    "source": [
     "identified_estimand = model.identify_effect(proceed_when_unidentifiable=True)\n",
-    "print(identified_estimand)"
+    "print(identified_estimand)\n",
+    "                                                                                                                                                                                                                                     \n",
+    "\n",
+    "from dowhy.causal_graph import CausalGraph\n",
+    "\n",
+    "graph = CausalGraph(\n",
+    "            data_dict[\"treatment_name\"],\n",
+    "            data_dict[\"outcome_name\"],\n",
+    "            common_cause_names=data_dict[\"common_causes_names\"],\n",
+    "            observed_node_names=df.columns.tolist(),\n",
+    ")\n",
+    "\n",
+    "nx_graph = graph._graph"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from dowhy.causal_identifier import identify_effect\n",
+    "\n",
+    "identified_estimand = identify_effect(nx_graph, action_nodes=data_dict[\"treatment_name\"], outcome_nodes=data_dict[\"outcome_name\"], observed_nodes=list(graph.get_all_nodes(include_unobserved=False)))"
    ]
   },
   {
@@ -197,7 +229,21 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "res_random=model.refute_estimate(identified_estimand, estimate, method_name=\"random_common_cause\")\n",
+    "from dowhy.causal_refuters import refute_estimate\n",
+    "from typing import List\n",
+    "\n",
+    "refuter_class = dowhy.causal_refuters.get_class_object(\"random_common_cause\")\n",
+    "\n",
+    "refuters = list()\n",
+    "refuters.append(refuter_class)\n",
+    "\n",
+    "res_random = refute_estimate(data=df, \n",
+    "                             treatment_name=data_dict[\"treatment_name\"], \n",
+    "                             target_estimand=identified_estimand,\n",
+    "                             outcome_name=data_dict[\"outcome_name\"], \n",
+    "                             identified_estimand=identified_estimand, \n",
+    "                             estimate=estimate,\n",
+    "                             refuters=refuters)\n",
     "print(res_random)"
    ]
   },
@@ -214,9 +260,20 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "res_placebo=model.refute_estimate(identified_estimand, estimate,\n",
-    "        method_name=\"placebo_treatment_refuter\", placebo_type=\"permute\")\n",
-    "print(res_placebo)"
+    "refuter_class = dowhy.causal_refuters.get_class_object(\"placebo_treatment_refuter\")\n",
+    "\n",
+    "refuters = list()\n",
+    "refuters.append(refuter_class)\n",
+    "\n",
+    "res_placebo = refute_estimate(data=df, \n",
+    "                             treatment_name=data_dict[\"treatment_name\"], \n",
+    "                             target_estimand=identified_estimand,\n",
+    "                             outcome_name=data_dict[\"outcome_name\"], \n",
+    "                             identified_estimand=identified_estimand, \n",
+    "                             estimate=estimate,\n",
+    "                             refuters=refuters,\n",
+    "                             placebo_type=\"permute\")\n",
+    "print(res_placebo)\n"
    ]
   },
   {
@@ -232,9 +289,20 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "res_subset=model.refute_estimate(identified_estimand, estimate,\n",
-    "        method_name=\"data_subset_refuter\", subset_fraction=0.9)\n",
-    "print(res_subset)\n"
+    "refuter_class = dowhy.causal_refuters.get_class_object(\"data_subset_refuter\")\n",
+    "\n",
+    "refuters = list()\n",
+    "refuters.append(refuter_class)\n",
+    "\n",
+    "res_subset = refute_estimate(data=df, \n",
+    "                             treatment_name=data_dict[\"treatment_name\"], \n",
+    "                             target_estimand=identified_estimand,\n",
+    "                             outcome_name=data_dict[\"outcome_name\"], \n",
+    "                             identified_estimand=identified_estimand, \n",
+    "                             estimate=estimate,\n",
+    "                             refuters=refuters,\n",
+    "                             subset_fraction=0.9)\n",
+    "print(res_subset)"
    ]
   },
   {
@@ -261,7 +329,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.8.5"
+   "version": "3.11.8"
   },
   "toc": {
    "base_numbering": 1,

diff --git a/dowhy/causal_refuters/refute_estimate.py b/dowhy/causal_refuters/refute_estimate.py
@@ -55,7 +55,9 @@ def refute_estimate(
 
     results = []
     for refuter in refuters:
+        print("REFUTER: ", refuter)
         refute = refuter(**refuter_kwargs)
+        print("REFUTE: ", refute)
         if isinstance(refute, list):
             results.extend(refute)
         else: