When training an XGBoost model for binary classification tasks, log loss (logloss) is a commonly used evaluation metric. Logloss measures the accuracy of the predicted probabilities and heavily penalizes confident misclassifications.
By setting eval_metric='logloss', you can monitor your model’s performance during training and enable early stopping to prevent overfitting.
Here’s an example of how to use logloss 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 binary classification dataset
X, y = make_classification(n_samples=1000, n_features=10, n_informative=5, n_redundant=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 logloss as the evaluation metric
model = XGBClassifier(n_estimators=100, eval_metric='logloss', 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)], verbose=False)
# Retrieve the logloss values from the training process
results = model.evals_result()
epochs = len(results['validation_0']['logloss'])
x_axis = range(0, epochs)
# Plot the logloss values
plt.figure()
plt.plot(x_axis, results['validation_0']['logloss'], label='Test')
plt.legend()
plt.xlabel('Number of Boosting Rounds')
plt.ylabel('Log Loss')
plt.title('XGBoost Log Loss Performance')
plt.show()
In this example, we generate a synthetic binary classification dataset using scikit-learn’s make_classification function. We then split the data into training and testing sets.
We create an instance of XGBClassifier and set eval_metric='logloss' to specify log loss as the evaluation metric. We also set early_stopping_rounds=10 to enable early stopping if the log loss 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 log loss values using the evals_result() method.
Finally, we plot the log loss 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 log loss as the evaluation metric, we can effectively monitor the model’s binary classification performance, prevent overfitting through early stopping, and select the best model based on the lowest log loss value.