Decision trees in Machine Learning are used for training machine learning models for classification or regression tasks. In general, we use the term overfitting or underfitting to evaluate the working and prediction mechanism. But we use a pruning technique to reduce the cost complexity of the model.
In the Decision Tree algorithm, there are decision nodes and edges. We can use pruning to cut less useful nodes, which will reduce the cost complexity of the decision tree. The pruning technique removes a subtree after calculating its importance for model accuracy.
A more deep insight will be provided if you get to learn Machine Learning Tutorial. Learning this will eventually clear your concepts based on Decision Trees.
Hope this answer will help you!