Merge pull request #109 from janpipek/stats_header

Stats header + median

HISTORY.txt

@@ -1,3 +1,12 @@
+0.7.2 (?? Jun 2024)
+-------------------
+- Improved: stats box in plots
+
+0.7.1 (6 Jun 2024)
+------------------
+- Fixed: Newer polars compatibility, deal with Nones
+- Added: .physt accessors for polars series and dataframes
+
 0.7.0 (31 May 2024)
 -------------------
 - Added: Support for pola.rs series/dataframes

src/physt/_construction.py

@@ -435,8 +435,10 @@ def calculate_1d_frequencies(
             raise ValueError(
                 f"Weights must have the same shape as data, {weights_array.shape} != {data_array.shape}"
             )
+        equal_weights = weights_array.max() - weights_array.min() == 0
     else:
         weights_array = np.ones_like(data_array, dtype=int)
+        equal_weights = True
 
     # Prepare dtype
     inferred_dtype: np.dtype = np.dtype(dtype or weights_array.dtype)
@@ -483,6 +485,8 @@ def calculate_1d_frequencies(
             min=float(data_array.min()),
             max=float(data_array.max()),
             weight=float(weights_array.sum()),
+            # TODO: Support median with weights?
+            median=np.median(data_array) if equal_weights else np.nan,
         )
     return frequencies, errors2, underflow, overflow, stats
 

src/physt/plotting/__init__.py

@@ -71,6 +71,8 @@
     If True, display a small box with statistical info
 stats_loc: 1 | 2 | 3 | 4 | "upper right" | "upper left" | "lower left" | "lower right"
     Where to display the stats box (similar to legend)
+stats_title:
+    What is displayed in the stats box header
 
 2D heatmaps
 ~~~~~~~~~~~

src/physt/plotting/matplotlib.py

@@ -984,6 +984,7 @@ def _add_stats_box(
     h1: Histogram1D,
     ax: Axes,
     stats: Union[str, bool, Collection[str]] = "all",
+    title: Optional[str] = None,
     loc: Union[int, str, None] = None,
 ) -> None:
     """Insert a small legend-like box with statistical information.
@@ -993,6 +994,7 @@ def _add_stats_box(
     stats : False | "all" or True | field | list of fields
         What info to display
     loc: As with legend(loc=...) but we do not support auto
+    title: What is displayed above the stats_box
 
     Note
     ----
@@ -1005,6 +1007,7 @@ def _add_stats_box(
         "underflow",
         "overflow",
         "std",
+        "median",
         "total",
     ]
 
@@ -1027,6 +1030,8 @@ def _add_stats_box(
         used_stats = list(stats)  # type: ignore
 
     text_frags = []
+    if title:
+        text_frags.append(title)
     if "total" in used_stats:
         text_frags.append(f"Total: {h1.total}")
     if "underflow" in used_stats:
@@ -1036,21 +1041,25 @@ def _add_stats_box(
         if h1.overflow:
             text_frags.append(f"Overflow: {h1.overflow}")
     if "mean" in used_stats:
-        mean = h1.mean()
+        mean = h1.statistics.mean()
         if mean is not None:
             text_frags.append(f"Mean: {mean:.2f}")
+    if "median" in used_stats:
+        median = h1.statistics.median
+        if median is not None:
+            text_frags.append(f"Median: {median:.2f}")
     if "std" in used_stats:
-        std = h1.std()
+        std = h1.statistics.std()
         if std is not None:
             text_frags.append(f"Std.dev: {std:.2f}")
     if "min" in used_stats:
-        min = h1.min()
-        if min is not None:
-            text_frags.append(f"Min.: {min:.2f}")
+        min_ = h1.statistics.min
+        if min_ is not None:
+            text_frags.append(f"Min.: {min_:.2f}")
     if "max" in used_stats:
-        max = h1.max()
-        if max is not None:
-            text_frags.append(f"Max.: {max:.2f}")
+        max_ = h1.statistics.max
+        if max_ is not None:
+            text_frags.append(f"Max.: {max_:.2f}")
 
     text = "\n".join(text_frags)
 

src/physt/statistics.py

@@ -15,6 +15,8 @@
 class Statistics:
     """Container of statistics accumulative data."""
 
+    # TODO: Reconsider changing mean, std to properties
+
     sum: float = 0.0
     """Weighted sum of all values entered into histogram."""
 
@@ -30,6 +32,8 @@ class Statistics:
     weight: float = 0.0
     """The total weight of values used to construct the histogram."""
 
+    median: float = np.nan
+
     def mean(self) -> float:
         """Statistical mean of all values entered into histogram (weighted)."""
         try:
@@ -63,6 +67,7 @@ def __add__(self, other: Any) -> Statistics:
             min=min(self.min, other.min),
             max=max(self.max, other.max),
             weight=self.weight + other.weight,
+            median=np.nan,
         )
 
     def __mul__(self, other: Any) -> Statistics:

tests/test_stats.py

@@ -65,6 +65,12 @@ def test_weight(self, histogram, use_weights):
         else:
             assert histogram.statistics.weight == 4
 
+    def test_median(self, histogram, use_weights):
+        if use_weights:
+            assert np.isnan(histogram.statistics.median)
+        else:
+            assert histogram.statistics.median == 2.5
+
     def test_mean(self, histogram, use_weights):
         if use_weights:
             assert histogram.statistics.mean() == 2.8