Decision tree

src: i.ytimg.com

Decision tree is a decision support tool that uses tree-like graphs or decision models and their possible consequences, including incidents, resource costs, and utilities. This is one way to display an algorithm that contains only conditional control statements.

Decision trees are typically used in operations research, particularly in decision analysis, to help identify the most likely strategies for achieving goals, but are also a popular tool in machine learning.

Video Decision tree

Ikhtisar

Decision tree is a flowchart-like structure in which each internal node is a "test" on the attribute (eg whether the coin flip appears head or tail), each branch is the result of the test, and each leaf node represents a class label (decision taken after counting all attributes). The path from root to leaf represents the classification rules.

In decision analysis, decision trees and influence diagrams are closely linked as visual and analytical decision support tools, where the expected value (or expected utility) of a competing alternative is calculated.

The decision tree consists of three types of nodes:

1. Results node - usually represented by a box
2. Chance nodes - usually represented by a circle
3. End netting - usually represented by triangle

Decision trees are commonly used in operations research and operations management. If, in practice, decisions must be taken online without recall under incomplete knowledge, decision trees should be paralleled by the probability model as the best choice model or online selection model algorithm. Another use of the decision tree is as a descriptive tool for calculating conditional probabilities.

Decision trees, influence diagrams, utility functions, and other decision analysis tools and tools are taught to undergraduate students in business schools, health economics, and public health, and are examples of research operations or management science methods.

Maps Decision tree

Decision tree creation block

Decision tree elements

Taken from left to right, the decision tree only has a bursting node (dividing path) but no sink node (converging path). Therefore, used manually, they can grow very large and often difficult to draw completely by hand. Traditionally, decision trees have been created manually - like the example on the side shows - though increasingly, special software is used.

Decision rule

Decision trees can be linearized into decision rules , where the result is the contents of the leaf node, and conditions along the path form a conjunction in the if clause. In general, rules have the form:

if condition1 and condition2 and condition3 then the result.

Decision rules can be generated by building association rules with target variables on the right. They can also indicate a temporal or causal relationship.

Decision tree using flowchart symbol

Generally decision trees are drawn using flowchart symbols because it is easier for many people to read and understand.

Example analysis

Analysis can consider the preferences of decision makers (eg, corporations) or utility functions, for example:

The basic interpretation in this situation is that firms prefer risk and reward B under the realistic coefficient of risk preference (greater than $ 400K - in the risk aversion range, firms need to model the third strategy, "Both A and B").

Influence chart

Most information in the decision tree can be represented more compactly as an influence diagram, focusing attention on issues and relationships between events.

src: www.sfs.uni-tuebingen.de

Induction of association rules

Decision trees can also be seen as generative models of induction rules from empirical data. The optimal decision tree is then defined as the tree that accounts for most of the data, while minimizing the number of levels (or "questions"). Some algorithms for generating such optimized trees have been created, such as ID3/4/5, CLS, ASSISTANT, and CART.

src: i.ytimg.com

Advantages and disadvantages

Among decision support tools, decision trees (and influence diagrams) have several advantages. Decision tree:

• Simple to understand and interpret. People can understand the decision tree model after a brief explanation.
• Have value even with little hard data. Important insights can be generated based on experts who explain the situation (alternatively, probability, and cost) and their preference for results.
• Allow adding new possible scenarios.
• Helps determine the worst, best, and expected scores for different scenarios.
• Use the white box model. If the results provided are provided by the model.
• Can be combined with other decision techniques.

Lack of decision tree:

• They are unstable, meaning that small changes in data can cause major changes in the optimal decision tree structure.
• They are often relatively inaccurate. Many other predictors perform better with similar data. This can be improved by replacing a single decision tree with a random decision tree forest, but random forest is not easily interpreted as a single decision tree.
• For data including categorical variables with different number of levels, the acquisition of information in the bias decision tree supports those attributes with more levels.
• Calculations can be very complicated, especially if many values ​​are uncertain and/or if many results are linked.

src: scalar.usc.edu

See also

src: www.tonibest.com

References

src: www.mathworks.com

External links

• Tutorials and Examples of Extensive Decision Tree
• Gallery of decision tree samples

Source of the article : Wikipedia