Advantages and Disadvantages of decision tree

There are advantages and disadvantages of decision trees and drawbacks to using decision trees, and they vary depending on the nature of the problem being solved. A decision tree is a graphical representation of a problem's possible answers under specific assumptions. A decision tree's structure is identical to that of other tree-based data structures, such as the balanced selection tree (BST), the binary tree (BT), and the alternating complete (AVL) tree. A decision tree can be made manually or with the help of a graphics editor or other software. In layman's terms, decision trees help keep the conversation focused during group deliberation.

Using a Decision Tree and Its Pros and Cons

The advantages and disadvantages of decision trees and drawbacks are outlined below:


It's applicable to classification and regression issues alike: Decision trees are effective in both regression and classification problems, allowing them to be utilised for the prediction of both continuous and discrete values.

Because decision trees are straightforward, they reduce the cognitive load associated with learning an algorithm.

They can be used to categorise data that does not follow a linear pattern.

Since the decision tree approach does not consider many weighted combinations simultaneously, it does not necessitate any change of the features when dealing with non-linear data.

When compared to KNN and other classification algorithms, they are much faster and more efficient.

Decision trees may process data of any kind, including numbers, categories, and booleans.

If using the Decision Tree, normalisation is unnecessary.

We don't have to worry about feature scaling in the decision tree, unlike in some other machine learning techniques. In addition, you can use random forests. The algorithms in question do not change their performance based on the size of the input.

We can use this information to better understand the value of certain characteristics.

Helpful for sifting through data: One of the quickest ways to zero in on the most important factors in a situation and the connections between them is to use a decision tree. The ability to generate additional variables/features for the output variable is enhanced by the use of decision trees.

There will be less of a need to clean up data: Since the presence of an outsider or the presence of incomplete data at a node of the decision tree does not change the outcome, the decision tree can function with less information.

Unlike traditional statistical methods, decision trees do not rely on any particular set of measurements. To avoid making arbitrary assumptions about spatial distribution and classifier structure, a non-parametric approach is used.


For millions of records with numeric variables, here is the decision tree split: Training a decision tree with two numerical variables is a time-consuming process since its complexity grows exponentially with the number of records.

Methods like random forests and XGBoost experience the same thing.

The many-featured decision tree: As the input grows, so does the intricacy of the training process, so be patient.

Acquiring wisdom from the training set's tree: Pre and post-processing overfit trimming, random forest ensemble learning.

The overfitting approach: When it comes to decision tree models, overfitting is one of the trickiest techniques to master. Placing limitations on the model's parameters and the pruning procedure helps alleviate the overfitting issue.

You are aware that overfitting data is typically required when developing a decision tree. The overfitting issue causes significant output variation, which in turn causes numerous estimation mistakes and, potentially, severely inaccurate results. Obtain complete lack of bias (overfitting), which increases variability.

The data in a decision tree can be reused, yet with slight changes the tree's result could be completely different. The use of techniques such as bagging and boosting can help reduce this decision tree variance.

No good can come of using it with large datasets: Larger datasets increase the likelihood that a single tree may grow a large number of nodes, which can increase complexity and lead to overfitting.

The optimal decision tree, which would be 100% efficient, may not be returned.