change Fed Stats output format (#2199)

* 1. change fed Stats output format : no need to change the format to fit the visualization needs
2. change the visualization to reformat the disctionary to fit the visualization needs

* formatting

examples/hello-world/step-by-step/higgs/stats/tabular_stats.ipynb

@@ -441,7 +441,9 @@
    "cell_type": "code",
    "execution_count": null,
    "id": "4af4d563-ec0e-4d03-a6ae-008d9ba62171",
-   "metadata": {},
+   "metadata": {
+    "tags": []
+   },
    "outputs": [],
    "source": [
     "!nvflare job create -w stats_df -force -j /tmp/nvflare/jobs/stats_df \\-sd code \\\n",
@@ -485,7 +487,9 @@
    "cell_type": "code",
    "execution_count": null,
    "id": "3fb4ffde-627c-4f72-b114-fb9e92c1dc6f",
-   "metadata": {},
+   "metadata": {
+    "tags": []
+   },
    "outputs": [],
    "source": [
     "!nvflare simulator /tmp/nvflare/jobs/stats_df -w /tmp/nvflare/tabular/stats_df -n 3 -t 3"
@@ -516,6 +520,18 @@
     "!ls -al /tmp/nvflare/tabular/stats_df/simulate_job/statistics/\n"
    ]
   },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "48c955f1-d002-4d5a-a408-6c9c78ad9854",
+   "metadata": {
+    "tags": []
+   },
+   "outputs": [],
+   "source": [
+    "!cat /tmp/nvflare/tabular/stats_df/simulate_job/statistics/stats.json"
+   ]
+  },
   {
    "cell_type": "markdown",
    "id": "653f83f8-f96f-4943-af27-c5e6551d3449",

nvflare/app_common/workflows/statistics_controller.py

@@ -384,34 +384,44 @@ def _combine_all_statistics(self):
         for statistic in filtered_client_statistics:
             for client in self.client_statistics[statistic]:
                 for ds in self.client_statistics[statistic][client]:
-                    client_dataset = f"{client}-{ds}"
                     for feature_name in self.client_statistics[statistic][client][ds]:
                         if feature_name not in result:
                             result[feature_name] = {}
                         if statistic not in result[feature_name]:
                             result[feature_name][statistic] = {}
 
+                        if client not in result[feature_name][statistic]:
+                            result[feature_name][statistic][client] = {}
+
+                        if ds not in result[feature_name][statistic][client]:
+                            result[feature_name][statistic][client][ds] = {}
+
                         if statistic == StC.STATS_HISTOGRAM:
                             hist: Histogram = self.client_statistics[statistic][client][ds][feature_name]
                             buckets = StatisticsController._apply_histogram_precision(hist.bins, self.precision)
-                            result[feature_name][statistic][client_dataset] = buckets
+                            result[feature_name][statistic][client][ds] = buckets
                         else:
-                            result[feature_name][statistic][client_dataset] = round(
+                            result[feature_name][statistic][client][ds] = round(
                                 self.client_statistics[statistic][client][ds][feature_name], self.precision
                             )
 
         precision = self.precision
         for statistic in filtered_global_statistics:
             for ds in self.global_statistics[statistic]:
-                global_dataset = f"{StC.GLOBAL}-{ds}"
                 for feature_name in self.global_statistics[statistic][ds]:
+                    if StC.GLOBAL not in result[feature_name][statistic]:
+                        result[feature_name][statistic][StC.GLOBAL] = {}
+
+                    if ds not in result[feature_name][statistic][StC.GLOBAL]:
+                        result[feature_name][statistic][StC.GLOBAL][ds] = {}
+
                     if statistic == StC.STATS_HISTOGRAM:
                         hist: Histogram = self.global_statistics[statistic][ds][feature_name]
                         buckets = StatisticsController._apply_histogram_precision(hist.bins, self.precision)
-                        result[feature_name][statistic][global_dataset] = buckets
+                        result[feature_name][statistic][StC.GLOBAL][ds] = buckets
                     else:
                         result[feature_name][statistic].update(
-                            {global_dataset: round(self.global_statistics[statistic][ds][feature_name], precision)}
+                            {StC.GLOBAL: {ds: round(self.global_statistics[statistic][ds][feature_name], precision)}}
                         )
 
         return result

nvflare/app_opt/statistics/visualization/statistics_visualization.py

@@ -17,6 +17,17 @@
 from nvflare.fuel.utils.import_utils import optional_import
 
 
+def convert_data(feature_metrics) -> dict:
+    converted = {}
+    for statistic in feature_metrics:
+        converted[statistic] = {}
+        for site in feature_metrics[statistic]:
+            for ds in feature_metrics[statistic][site]:
+                site_dataset = f"{site}-{ds}"
+                converted[statistic][site_dataset] = feature_metrics[statistic][site][ds]
+    return converted
+
+
 class Visualization:
     def import_modules(self):
         display, import_flag = optional_import(module="IPython.display", name="display")
@@ -27,17 +38,23 @@ def import_modules(self):
             print(pd.failure)
         return display, pd
 
-    def show_stats(self, data, white_list_features=[]):
+    def show_stats(self, data, white_list_features=None):
+        if white_list_features is None:
+            white_list_features = []
+
         display, pd = self.import_modules()
         all_features = [k for k in data]
         target_features = self._get_target_features(all_features, white_list_features)
         for feature in target_features:
             print(f"\n{feature}\n")
             feature_metrics = data[feature]
-            df = pd.DataFrame.from_dict(feature_metrics)
+            converted = convert_data(feature_metrics)
+            df = pd.DataFrame.from_dict(converted)
             display(df)
 
-    def show_histograms(self, data, display_format="sample_count", white_list_features=[], plot_type="both"):
+    def show_histograms(self, data, display_format="sample_count", white_list_features=None, plot_type="both"):
+        if white_list_features is None:
+            white_list_features = []
         feature_dfs = self.get_histogram_dataframes(data, display_format, white_list_features)
         self.show_dataframe_plots(feature_dfs, plot_type)
 
@@ -54,7 +71,9 @@ def show_dataframe_plots(self, feature_dfs, plot_type="both"):
             else:
                 print(f"not supported plot type: '{plot_type}'")
 
-    def get_histogram_dataframes(self, data, display_format="sample_count", white_list_features=[]) -> Dict:
+    def get_histogram_dataframes(self, data, display_format="sample_count", white_list_features=None) -> Dict:
+        if white_list_features is None:
+            white_list_features = []
         display, pd = self.import_modules()
         (hists, edges) = self._prepare_histogram_data(data, display_format, white_list_features)
         all_features = [k for k in edges]
@@ -69,15 +88,18 @@ def get_histogram_dataframes(self, data, display_format="sample_count", white_li
 
         return feature_dfs
 
-    def _prepare_histogram_data(self, data, display_format="sample_count", white_list_features=[]):
+    def _prepare_histogram_data(self, data, display_format="sample_count", white_list_features=None):
+        if white_list_features is None:
+            white_list_features = []
         all_features = [k for k in data]
         target_features = self._get_target_features(all_features, white_list_features)
 
         feature_hists = {}
         feature_edges = {}
 
         for feature in target_features:
-            xs = data[feature]["histogram"]
+            converted = convert_data(data[feature])
+            xs = converted["histogram"]
             hists = {}
             feature_edges[feature] = []
             for i, ds in enumerate(xs):
@@ -103,7 +125,10 @@ def sum_counts_in_histogram(self, hist):
             sum_value += bucket[2]
         return sum_value
 
-    def _get_target_features(self, all_features, white_list_features=[]):
+    def _get_target_features(self, all_features, white_list_features=None):
+        if white_list_features is None:
+            white_list_features = []
+
         target_features = white_list_features
         if not white_list_features:
             target_features = all_features