Added download of synthetic aggregates (#91)

Co-authored-by: Darren Edge <darren.edge@example.com>
microsoft · Feb 7, 2025 · 9124483 · 9124483
1 parent 28f600f
commit 9124483
Show file tree

Hide file tree

Showing 3 changed files with 31 additions and 6 deletions.
diff --git a/app/workflows/anonymize_case_data/variables.py b/app/workflows/anonymize_case_data/variables.py
@@ -24,6 +24,7 @@ def create_session(self, prefix):
         self.anonymize_last_synthetic_file_name = SessionVariable("", prefix)
         self.anonymize_last_aggregate_file_name = SessionVariable("", prefix)
         self.anonymize_synthetic_df = SessionVariable(pd.DataFrame(), prefix)
+        self.anonymize_synthetic_aggregate_df = SessionVariable(pd.DataFrame(), prefix)
         self.anonymize_aggregate_df = SessionVariable(pd.DataFrame(), prefix)
         self.anonymize_epsilon = SessionVariable(12.0, prefix)
         # self.anonymize_sen_agg_rep = SessionVariable(pd.DataFrame(), prefix)

diff --git a/app/workflows/anonymize_case_data/workflow.py b/app/workflows/anonymize_case_data/workflow.py
@@ -142,6 +142,7 @@ def create(sv: ds_variables.SessionVariables, workflow: None):
                             )
                             sv.anonymize_synthetic_df.value = acd.synthetic_df
                             sv.anonymize_aggregate_df.value = acd.aggregate_df
+                            sv.anonymize_synthetic_aggregate_df.value = acd.synthetic_aggregate_df
                             sv.anonymize_delta.value = f"{acd.delta:.2e}"
                 if epsilon > 12:
                     st.warning(
@@ -209,12 +210,21 @@ def create(sv: ds_variables.SessionVariables, workflow: None):
                         use_container_width=True,
                         height=700,
                     )
-                    st.download_button(
-                        "Download Synthetic data",
-                        data=sv.anonymize_synthetic_df.value.to_csv(index=False),
-                        file_name="synthetic_data.csv",
-                        mime="text/csv",
-                    )
+                    ca, cb = st.columns([4,5])
+                    with ca:
+                        st.download_button(
+                            "Download Synthetic data",
+                            data=sv.anonymize_synthetic_df.value.to_csv(index=False),
+                            file_name="synthetic_data.csv",
+                            mime="text/csv",
+                        )
+                    with cb:
+                        st.download_button(
+                            "Download Synthetic aggregates",
+                            data=sv.anonymize_synthetic_aggregate_df.value.to_csv(index=False),
+                            file_name="synthetic_aggregate_data.csv",
+                            mime="text/csv",
+                        )
 
     with queries_tab:
         if (

diff --git a/intelligence_toolkit/anonymize_case_data/api.py b/intelligence_toolkit/anonymize_case_data/api.py
@@ -36,6 +36,7 @@ def __init__(self) -> None:
         self.delta = 0
         self.sensitive_df = pd.DataFrame()
         self.aggregate_df = pd.DataFrame()
+        self.synthetic_aggregate_df = pd.DataFrame()
         self.synthetic_df = pd.DataFrame()
         self.aggregate_error_report = pd.DataFrame()
         self.synthetic_error_report = pd.DataFrame()
@@ -175,6 +176,19 @@ def anonymize_case_data(
         self.aggregate_df = self.aggregate_df.sort_values(
             by=["protected_count"], ascending=False
         )
+
+        self.synthetic_aggregate_df = pd.DataFrame(
+            data=synthetic_aggregates.items(),
+            columns=["selections", "protected_count"],
+        )
+        self.synthetic_aggregate_df.loc[len(self.synthetic_aggregate_df)] = [
+            "record_count",
+            self.protected_number_of_records,
+        ]
+        self.synthetic_aggregate_df = self.synthetic_aggregate_df.sort_values(
+            by=["protected_count"], ascending=False
+        )
+
         self.aggregate_error_report = ErrorReport(
             sensitive_aggregates_parsed, dp_aggregates_parsed
         ).gen()