minor changes

README.md

@@ -15,22 +15,49 @@ The **unravelsports** package aims to aid researchers, analysts and enthusiasts
 🌀 Features
 -----
 
-- ⚽ Converting **positional soccer data** into graphs to train **graph neural networks** by leveraging the powerful [**Kloppy**](https://github.com/PySport/kloppy/tree/master) data conversion standard and [**Spektral**](https://github.com/danielegrattarola/spektral) - a flexible framework for creating graph neural networks. 
-- ⚽ Randomizing and splitting data into **train, test and validation sets** along matches, sequences or possessions to avoid leakage and improve model quality.
-- ⚽ Due to the power of **Kloppy**, **unravelsports** supports these actions for _Metrica_, _Sportec_, _Tracab (CyronHego)_, _SecondSpectrum_, _SkillCorner_ and _StatsPerform_ tracking data.
+<ul style="list-style: none; padding: 0; margin-left: 1.2em;">
+  <li style="margin-bottom: 8px;">
+    <span style="display: inline-block; width: 1.2em; margin-right: 0.5em;">⚽</span>
+    Convert <strong>positional soccer data</strong> into graphs to train <strong>graph neural networks</strong> by leveraging the powerful <a href="https://github.com/PySport/kloppy/tree/master"><strong>Kloppy</strong></a> data conversion standard and <a href="https://github.com/danielegrattarola/spektral"><strong>Spektral</strong></a> - a flexible framework for creating GNNs.
+  </li>
+  <li style="margin-bottom: 8px;">
+    <span style="display: inline-block; width: 1.2em; margin-right: 0.5em;">⚽</span>
+    Randomize and split data into <strong>train, test and validation sets</strong> along matches, sequences or possessions to avoid leakage and improve model quality.
+  </li>
+  <li style="margin-bottom: 8px;">
+    <span style="display: inline-block; width: 1.2em; margin-right: 0.5em;">⚽</span>
+    Due to the power of <strong>Kloppy</strong>, <strong>unravelsports</strong> supports these actions for <em>Metrica</em>, <em>Sportec</em>, <em>Tracab (CyronHego)</em>, <em>SecondSpectrum</em>, <em>SkillCorner</em> and <em>StatsPerform</em> tracking data.
+  </li>
+  <li style="margin-bottom: 8px;">
+    <span style="display: inline-block; width: 1.2em; margin-right: 0.5em;">⏳</span>
+    <strong>More to come...</strong>
+  </li>
+</ul>
 
 🌀 Getting Started
 -----
-📖 The [**Getting Started Jupyter Notebook**](examples/0_getting_started.ipynb) explains how to convert any positional tracking data from **Kloppy** to **Spektral GNN** in a few easy steps while walking you through the most important features and documentation.
-
-📖 The [**Graph Converter Tutorial Jupyter Notebook**](examples/1_tutorial_graph_converter.ipynb) gives an in-depth walkthrough.
+<ul style="list-style: none; padding: 0; margin-left: 1.2em;">
+  <li style="margin-bottom: 8px;">
+    <span style="display: inline-block; width: 1.2em; margin-right: 0.5em;">📖</span>
+    The <a href="examples/0_getting_started.ipynb"><strong>Getting Started Jupyter Notebook</strong></a> explains how to convert any positional tracking data from <strong>Kloppy</strong> to <strong>Spektral GNN</strong> in a few easy steps while walking you through the most important functionality.
+  </li>
+  <li style="margin-bottom: 8px;">
+    <span style="display: inline-block; width: 1.2em; margin-right: 0.5em;">📖</span>
+    The <a href="examples/1_tutorial_graph_converter.ipynb"><strong>Graph Converter Tutorial Jupyter Notebook</strong></a> gives an in-depth walkthrough.
+  </li>
+</ul>
 
 🌀 Documentation
 -----
 For now, follow the [**Graph Converter Tutorial**](examples/1_tutorial_graph_converter.ipynb), more documentation will follow!
 
 Additional reading:
-- 📖 [A Graph Neural Network Deep-dive into Successful Counterattacks {A. Sahasrabudhe & J. Bekkers, 2023}](https://github.com/USSoccerFederation/ussf_ssac_23_soccer_gnn/tree/main)
+<ul style="list-style: none; padding: 0; margin-left: 1.2em;">
+  <li style="margin-bottom: 8px;">
+    <span style="display: inline-block; width: 1.2em; margin-right: 0.5em;">📖</span>
+    <a href="https://github.com/USSoccerFederation/ussf_ssac_23_soccer_gnn/tree/main"><strong>A Graph Neural Network Deep-dive into Successful Counterattacks {A. Sahasrabudhe & J. Bekkers, 2023}</strong></a>
+  </li>
+</ul>
 
 🌀 Installation
 ----
@@ -44,7 +71,7 @@ pip install unravelsports
 
 🌀 Contributing
 ----
-All contributions, bug reports, bug fixes, documentation improvements, enhancements, and ideas are welcome.
+All contributions, bug reports, bug fixes, documentation improvements, enhancements, and ideas are welcome. Feel free to create a Pull Request for any improvements you make that do not contribute to winning more games!
 
 An overview on how to contribute can be found in the [**contributing guide**](CONTRIBUTING.md).
 

tests/test_kloppy.py

@@ -117,17 +117,19 @@ def test_conversion(self, gnnc: GraphConverter):
         assert data.orientation == Orientation.STATIC_HOME_AWAY
         assert data.attacking_players == data.home_players
         assert data.defending_players == data.away_players
-        
+
         hp = data.home_players[3]
-        assert -19.582426479899993 == pytest.approx(hp.x1, abs=1e-5) 
-        assert 24.3039460863 == pytest.approx(hp.y1, abs=1e-5) 
-        assert -19.6022318885 == pytest.approx(hp.x2, abs=1e-5) 
-        assert 24.1632567814 == pytest.approx(hp.y2, abs=1e-5) 
-        assert hp.position.shape == (2, )
-        np.testing.assert_allclose(hp.position, np.asarray([hp.x1, hp.y1]), rtol=1e-4, atol=1e-4)
+        assert -19.582426479899993 == pytest.approx(hp.x1, abs=1e-5)
+        assert 24.3039460863 == pytest.approx(hp.y1, abs=1e-5)
+        assert -19.6022318885 == pytest.approx(hp.x2, abs=1e-5)
+        assert 24.1632567814 == pytest.approx(hp.y2, abs=1e-5)
+        assert hp.position.shape == (2,)
+        np.testing.assert_allclose(
+            hp.position, np.asarray([hp.x1, hp.y1]), rtol=1e-4, atol=1e-4
+        )
         assert hp.is_gk == False
         assert hp.next_position[0] - hp.position[0]
-                
+
         assert data.ball_carrier_idx == 1
         assert len(data.home_players) == 6
         assert len(data.away_players) == 4

unravel/utils/objects/default_ball.py

@@ -24,8 +24,10 @@ def __post_init__(self):
 
     def set_velocity(self):
         delta_time = 1.0 / self.fps
-
-        if not (np.any(np.isnan(self.next_position3D)) or np.any(np.isnan(self.position3D))):
+
+        if not (
+            np.any(np.isnan(self.next_position3D)) or np.any(np.isnan(self.position3D))
+        ):
             vx = (self.next_position3D[0] - self.position3D[0]) / delta_time
             vy = (self.next_position3D[1] - self.position3D[1]) / delta_time
             vz = (self.next_position3D[2] - self.position3D[2]) / delta_time
@@ -43,7 +45,6 @@ def set_velocity(self):
 
         self.speed = np.sqrt(vx**2 + vy**2 + vz**2)
 
-
     def invert_position(self):
         self.next_position = self.next_position * -1.0
         self.position = self.position * -1.0

unravel/utils/objects/default_player.py

@@ -22,7 +22,6 @@ class DefaultPlayer(object):
     )  # velocity vector
     speed: float = 0.0  # actual speed in m/s
     is_gk: bool = False
-
 
     def __post_init__(self):
         self.next_position = np.asarray([self.x2, self.y2], dtype=float)
@@ -42,17 +41,19 @@ def invert_position(self):
 
     def set_velocity(self):
         dt = 1.0 / self.fps
-        if not (np.any(np.isnan(self.next_position)) or np.any(np.isnan(self.position))):
+        if not (
+            np.any(np.isnan(self.next_position)) or np.any(np.isnan(self.position))
+        ):
             vx = (self.next_position[0] - self.position[0]) / dt
             vy = (self.next_position[1] - self.position[1]) / dt
         else:
             vx = 0
             vy = 0
 
         self.velocity = np.asarray([vx, vy], dtype=float)
-        
+
         # Re-check if any component of velocity is NaN and set to zero if it is
         if np.any(np.isnan(self.velocity)):
             self.velocity = np.asarray([0.0, 0.0], dtype=float)
-            
+
         self.speed = np.sqrt(vx**2 + vy**2)

unravel/utils/objects/default_tracking.py

@@ -35,7 +35,6 @@ class DefaultTrackingModel:
     ball_carrier_treshold: bool = 25.0
     verbose: bool = False
     pad_n_players: bool = None
-
 
     def __post_init__(self):
         self.home_players: List[DefaultPlayer] = list()
@@ -207,7 +206,7 @@ def set_objects_from_frame(
                         y1=coords.y,
                         y2=next_coords.y,
                         is_visible=True,
-                        fps=self.fps
+                        fps=self.fps,
                     )
 
                     if pid.team.ground == Ground.HOME:

unravel/utils/utils.py

@@ -23,7 +23,7 @@ def dummy_graph_ids(dataset: TrackingDataset) -> Dict:
     """
     if not isinstance(dataset, TrackingDataset):
         raise TypeError("dataset should be of type TrackingDataset (from kloppy)")
-    
+
     from uuid import uuid4
 
     graph_ids = dict()