Decision tree and rule engine, decision tree Rule Engine

Decision tree and rule engine, decision tree Rule Engine

Use of Decision Trees

Decision Trees are often used in data mining and are one of the most basic algorithms. Almost everyone who has learned Data Mining knows decision trees. However, the original use of the restored decision tree is more practical and intuitive when used for decision making or decision-making. Its tree structure guides people to focus on several of the most important directions when facing a decision. These directions are fixed and further subdivided. In recent years, graphics/office automation tools such as mind-oriented graphs have gradually emerged and are well received by everyone, that is, a good implementation of decision trees. However, decision trees are not very commonly used in various enterprise application systems. In the final analysis, decision trees are thought-oriented content and erratic content, making it very difficult to form structured content. In addition, most business systems on the market use relational databases, which are very useful in processing formatted data, but tree data processing is not qualitative. Therefore, some technical companies are gradually Using Object-based databases. On the other hand, decisions and judgments in decision trees are relatively irregular. Many content is more like programming by programmers. They are rules rather than information, which leads to difficulties in traditional business systems. Composition and Program Performance of Decision TreesThe decision tree uses a tree structure to express business rules. As shown in. Each non-leaf node code a decision/decision, and the leaf node executes the action. Each edge represents an optional decision value, which can be understood as a judgment. For example, A = red or = blue is an optional value, and B is A decision node.
However, in program implementation, it is not necessarily such an organization. Generally, the node text is easier to see than the online text, in addition, various programming languages are based on tree-type controls, which are expressed in the form of nodes and seldom use connection lines for tables. Therefore, the most common representation in a program is to concentrate a large amount of information in the results. For example, if the node A = red is written, A = red is written instead of another name. For the leaf node, You need to display the Action in more detail ), the effect may be as follows:
The data of some decision tables can also be organized as decision trees or used to express more suitable data. The decision tree can quickly and effectively associate multiple related rules. The decision logic at each level may be clearly viewed through the tree relationship. When performing an operation, you can quickly traverse each decision node and check whether the operation meets the conditions. If the operation meets the conditions, you can traverse it down. Finally, find the applicable conditions and applicable operation actions. Parking lot pricing example using decision treeThe Business System caller does not need to fill in any code, and all calculations are performed in CKRule. The settings in CKRule are as follows. Park1_pf = new park1_(); _ pf. parkType = cmbParkType. text; _ pf. distType = cmbDistType. text; _ pf. cardType = cmbCardType. text; _ pf. partTime = Convert. toDouble (numericUpDown1.Value); _ pf = new RuleFacade (). exec ("cost-parking fee calculation-Decision Tree", _ pf); txt133. text = _ pf. cost + ""; to view the rule settings, use the CKRule editor to open "cost-parking Fee calculation-decision tree. ckp file, find the decision tree and master rule to view. Related source code, Demo download: http://www.ckrule.com/cn/demo.html
The decision tree method is

The decision tree classification algorithm is a prediction model used in data mining technology. It derives the decision tree representation form from a sample dataset without rules and is used for classification of the target dataset. It can be used to process high-dimensional data with good accuracy. Its construction does not require any domain knowledge or parameter settings, so it is suitable for probe-based Knowledge Discovery. At present, the decision tree classification algorithm has been successfully applied to classification in many fields, such as business, medicine, manufacturing and production, financial analysis, astronomy and molecular biology.
The overall structure of the decision tree is similar to the tree structure of the flowchart. Each internal node (non-leaf node) indicates a test on a certain attribute, and each branch indicates a test output, each leaf node (or endpoint) stores a class label. The top-level node of the tree is the root node.
In the process of building a decision tree, the top-down recursion method is used to perform attribute values (prediction variables) at the nodes in the tree) to avoid the complexity and complexity of decision trees and prevent the occurrence of over-fitting, the decision tree must be trimmed during or after the decision tree is generated.
"How to Use Decision Tree Classification ?" We can directly use the generated decision tree model, that is, to test the dataset attribute value on the decision tree, given an X, which has the same attributes as the sample data but has an unknown class number, based on a path from the root node to the leaf node, the leaf node stores the class prediction of the tuples. We can also convert the decision tree model into a classification rule set, each rule corresponds to the judgment condition from the root to the leaf node path and the category in the leaf node. Then, the rule set is used to classify unknown tuples.
Currently, the main decision tree classification algorithms include: In the late 1980s s and the early 1980s S, machine learning researcher I. Ross. Quinlan developed the decision tree algorithm called ID3. Quinlan later proposed C4.5. In 1984, several statisticians (L. Breiman, J. Friedman, R. Qlshen and C. Stone) published the CART ). C5.0 is a new algorithm introduced by Quinlan Based on The C4.5 algorithm. However, the C5.0 algorithm proposed by Quinlan is directly made into application software for commercialization, therefore, the steps and mathematical descriptions of the C5.0 algorithm by Quinlan have not yet been published. In addition, there is also the GLC tree developed by Zhang xiaohe for remote sensing image classification.

What is a decision tree?

Is an analysis technology. A decision tree is used to graphically describe a decision under consideration and the potential consequences of choosing this or that option. It will be used in the future when the consequences of certain scenarios or actions are uncertain. It comprehensively considers related probabilities and gains and losses for each logical path composed of events and decisions, and uses the expected monetary value analysis to help organizations identify the relative values of various alternatives.