def combined_symmetry_train_classifier(
data=None,
data_backup_filename='ShapeAndTextureScores',
filename_to_save_model='ShapeAndTextureModel'):
"""Train a random forest classifier with data from the "shapeAndTextureScores.csv" (created by merging shape scores
and texture scores) file following the expert diagnosis about symmetry (PH2 Dataset).
# Arguments :
data: As returned by the texture_symmetry_scores function (optional).
backup_filename: Only if data is None, file to load data from.
filename_to_save_model: String or None.
# Outputs :
clf: The fitted classifier.
acc: The accuracy score of the classifier
"""
if data is None:
data = pd.read_csv(
f"{package_path()}/data/patchesDataSet/{data_backup_filename}.csv",
index_col=False)
features = list(data)
del features[0]
else:
features = list(data)
del features[0] # Delete labels too
trainX = data[features][50:]
trainy = data.Labels[50:]
valX = data[features][:50]
valy = data.Labels[:50]
clf = RandomForestClassifier(n_estimators=10,
max_leaf_nodes=3,
random_state=2)
clf.fit(trainX, trainy)
preds = clf.predict(valX)
acc = accuracy_score(valy, preds)
if filename_to_save_model:
save_model(clf, filename_to_save_model)
return clf, acc
def classifierTrainer(maxLeafNodes,
data=None,
data_backup_file='patchesDataSet/Features',
filename_to_save_model='PatchClassifierModel'):
"""Train a random forest classifier with data from the patchesDataSet.
# Arguments :
maxLeafNodes: Int or None. Grow trees with max_leaf_nodes in best-first fashion. Best nodes are
defined as relative reduction in impurity. If None then unlimited number of leaf nodes (scikit-learn
RandomForestClassifier() documentation).
# Outputs :
clf: The fitted classifier.
acc: The accuracy score of the classifier
"""
if data is None:
data = pd.read_csv(
f"{package_path()}/data/patchesDataSet/{data_backup_file}.csv",
index_col=False)
features = list(data)
del features[0]
else:
features = list(data)
del features[-1] # Remove `Result` colname.
trainX = data[features][500:]
trainy = data.Result[500:]
valX = data[features][:500]
valy = data.Result[:500]
clf = RandomForestClassifier(max_leaf_nodes=maxLeafNodes, random_state=2)
clf.fit(trainX, trainy)
if filename_to_save_model:
save_model(clf, filename_to_save_model)
preds = clf.predict(valX)
acc = accuracy_score(valy, preds)
return clf, acc
def example():
"""Usage example of the main functionalities within this file. """
dataExtractorForTraining(10, 199, 4)
classifier, accScore = classifierTrainer(200)
print(f'Accuracy score: {accScore}')
save_model(classifier, "SimilarPatchClassifier")