gitea_admin/sientia-dataops-model-manager
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

SIENTIAPDE-1430: Refactor DataPreprocessor date filtering, make rce_train radius optional, and introduce constants for model names.

Browse Source
This commit is contained in:
Bruno Domingues
2025-12-19 12:28:07 -03:00
parent 9a77257920
commit 7cbb7022e5
3 changed files with 50 additions and 38 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
model_manager/sientia/metrics.py
View File

@@ -45,13 +45,13 @@ def silverman_radius(data: np.ndarray) -> float:
return radius
def rce_train(training_set: pd.DataFrame, radius: float) -> pd.DataFrame:
def rce_train(training_set: pd.DataFrame, radius: float | None = None) -> pd.DataFrame:
"""
Get the Reduced Coulomb Energy (RCE) prototypes.
Args:
training_set (pd.DataFrame): The training set
radius (float): The radius of the RCE prototypes
radius (float | None): The radius of the RCE prototypes. If None, computed using Silverman's rule.
Returns:
pd.DataFrame: The RCE prototypes
@@ -62,8 +62,8 @@ def rce_train(training_set: pd.DataFrame, radius: float) -> pd.DataFrame:
diff_vectors = train_vectors[:, np.newaxis] - train_vectors[np.newaxis, :]
distances = np.linalg.norm(diff_vectors, axis=-1)
# Non-parametric radius: Silverman Radius
radius = silverman_radius(distances.flatten())
# Non-parametric radius: Silverman Radius (compute if not provided)
effective_radius = radius if radius is not None else silverman_radius(distances.flatten())
# Initialize prototypes with the first vector
prototypes = [train_vectors[0]]
@@ -73,7 +73,7 @@ def rce_train(training_set: pd.DataFrame, radius: float) -> pd.DataFrame:
distances_to_prototypes = np.linalg.norm(prototypes - vector, axis=1)
# If no prototype is close, add the current vector as a new prototype
if np.all(distances_to_prototypes > radius):
if np.all(distances_to_prototypes > effective_radius):
prototypes.append(vector)
return pd.DataFrame(prototypes)