Bagging, short for Bootstrap Aggregating, is a machine learning ensemble technique that improves the accuracy of models. It works by creating multiple subsets of the training data through random sampling with replacement. Each subset is used to train a separate model, and the final prediction is made by averaging the outputs of all models for regression tasks or by majority voting for classification tasks. This method helps reduce overfitting, as it combines the strengths of various models while minimizing their individual weaknesses. Bagging is commonly used with algorithms like Decision Trees, leading to more robust and reliable predictions in various applications.