Minor modif: code which is useful to display results.

as_seg/scripts/display_results.py

+"""
+Script to display experimental results in a nice format.
+
+I should probably use scikit-learn's GridSearchCV, TODO.
+"""
+
+import pandas as pd
+from IPython.display import display
+import numpy as np
+import itertools
+
+def display_experimental_results(data, conditions, metrics, emphasis = None):
+    assert data.shape[-1] == len(metrics), f"Wrong number of metrics: {data.shape[-1]} != {len(metrics)}"
+    if type(conditions) == list:
+        conditions = np.array(conditions, dtype=object)
+    if len(conditions.shape) == 1 and type(conditions[0]) not in [list, tuple, np.ndarray]:
+        indexes = conditions
+    elif len(conditions.shape) > 1 or (len(conditions.shape) == 1 and type(conditions) in [list, tuple, np.ndarray]):
+        indexes = np.array(list(itertools.product(*conditions))) # cartesian_product
+    else:
+        raise NotImplementedError(f"Wrong format of conditions: {conditions}")
+    
+    dataframe = pd.DataFrame(data, columns = metrics, index = indexes)
+    if emphasis is not None:
+        display(dataframe.style.bar(subset=emphasis, color='#5fba7d'))
+    else:
+        display(dataframe)
+    
+def find_best_condition(data, conditions):
+    if type(conditions) == list:
+        conditions = np.array(conditions, dtype=object)
+    if len(conditions.shape) == 1 and type(conditions[0]) not in [list, tuple, np.ndarray]:
+        best_arg = np.argmax(data)
+        best_val = conditions[best_arg]
+        return best_val
+    elif len(conditions.shape) > 1 or (len(conditions.shape) == 1 and type(conditions) in [list, tuple, np.ndarray]):
+        best_arg_flatten = np.argmax(data)
+        best_arg_tuple = np.unravel_index(best_arg_flatten, data.shape)
+        best_vals = []
+        for idx, best_arg in enumerate(best_arg_tuple):
+            best_vals.append(conditions[idx][best_arg])
+        return np.array(best_vals)
+
+        
+