When training an XGBoost model for multi-class classification tasks, the Multi-Class Log Loss (mlogloss) is a commonly used evaluation metric. Mlogloss measures the dissimilarity between the predicted class probabilities and the actual class labels, with lower values indicating better performance.
By setting eval_metric='mlogloss', you can monitor your model’s performance during training and enable early stopping to prevent overfitting. Here’s an example of how to use mlogloss as the evaluation metric with XGBoost and scikit-learn:
from sklearn.datasets import make_classification
from sklearn.model_selection import train_test_split
from xgboost import XGBClassifier
import matplotlib.pyplot as plt
# Generate a synthetic multi-class classification dataset
X, y = make_classification(n_samples=1000, n_classes=3, n_features=10, n_informative=5, random_state=42)
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
# Create an XGBClassifier with mlogloss as the evaluation metric
model = XGBClassifier(n_estimators=100, eval_metric='mlogloss', early_stopping_rounds=10, random_state=42)
# Train the model with early stopping
model.fit(X_train, y_train, eval_set=[(X_test, y_test)])
# Retrieve the mlogloss values from the training process
results = model.evals_result()
epochs = len(results['validation_0']['mlogloss'])
x_axis = range(0, epochs)
# Plot the mlogloss values
plt.figure()
plt.plot(x_axis, results['validation_0']['mlogloss'], label='Test')
plt.legend()
plt.xlabel('Number of Boosting Rounds')
plt.ylabel('Multi-Class Log Loss')
plt.title('XGBoost Multi-Class Log Loss Performance')
plt.show()
In this example, we generate a synthetic multi-class classification dataset using scikit-learn’s make_classification function with 3 classes. We then split the data into training and testing sets.
We create an instance of XGBClassifier and set eval_metric='mlogloss' to specify mlogloss as the evaluation metric. We also set early_stopping_rounds=10 to enable early stopping if the mlogloss doesn’t improve for 10 consecutive rounds.
During training, we pass the testing set as the eval_set to monitor the model’s performance on unseen data. After training, we retrieve the mlogloss values using the evals_result() method.
Finally, we plot the mlogloss values against the number of boosting rounds to visualize the model’s performance during training. This plot helps us assess whether the model is overfitting or underfitting and determines the optimal number of boosting rounds.
By using mlogloss as the evaluation metric, we can effectively monitor the model’s multi-class classification performance, prevent overfitting through early stopping, and select the best model based on the lowest mlogloss value.