feat: entropy profile

popmon/analysis/profiling/hist_profiler.py

@@ -102,6 +102,7 @@ def _profile_1d_histogram(self, name, hist):
 
         # difference between htype=None and htype="all" are arguments (bin labels vs hist)
         profile.update(Profiles.run([hist], dim=1, htype=None))
+        profile.update(Profiles.run([hist], dim=-1, htype=None))
 
         # postprocessing TS
         if is_ts:
@@ -110,9 +111,6 @@ def _profile_1d_histogram(self, name, hist):
                 for k, v in profile.items()
             }
 
-        # postprocessing sum
-        profile["count"] = profile["filled"] + profile["nan"]
-
         return profile
 
     def _profile_nd_histogram(self, name, hist, dim):

popmon/analysis/profiling/profiles.py

@@ -130,6 +130,18 @@ def profile_count(hist):
     return int(sum_entries(hist))
 
 
+@Profiles.register(
+    key="entropy",
+    description="Entropy in nats",
+    dim=-1,
+    htype=None,
+)
+def profile_entropy(hist):
+    h = hist.bin_entries()
+    h = h / h.sum()
+    return -(h * np.ma.log(h)).sum()
+
+
 @Profiles.register(
     key="filled",
     description="Number of non-missing entries (non-NaN)",