CART-based decision-making tree construction method in cognitive computation

Abstract

The invention discloses a CART-based decision-making tree construction method in cognitive computation. The method comprises the following steps: initializing a root node of a decision-making tree and a characteristic attribute set corresponding to the root node according to establishment rules of the decision-making tree; carrying out sorting of data in a training set; judging whether all sample data in a node is of the same type or not; calculating the optimum characteristic attributes and the suboptimum characteristic attributes of the node to be split; judging whether the node satisfies the selected optimum splitting attributes and splitting conditions of an interruption mechanism or not; if the node satisfies the selected optimum splitting attributes and the splitting conditions of the interruption mechanism, carrying out splitting through the optimum splitting attributes, then iteratively replacing a current node with the node split through the characteristic attributes, and adding two new leaf nodes respectively from a left branch and a right branch, so as to achieve automatic splitting of the decision-making tree; and if the node does not satisfy the selected optimum splitting attributes or the splitting conditions of the interruption mechanism, waiting for data stream inputs, carrying out sample updating, and continuing to calculate node splitting. The method provided by the invention has the advantages that the accuracy of data stream processing is further improved, so that the possibility of system blocking is reduced.

Description

technical field [0001] The invention relates to the technical field of big data processing, in particular to a CART-based decision tree construction method in cognitive computing. Background technique [0002] The rapid development of emerging technologies such as cloud computing and the Internet of Things has caused the scale of data to grow at an unprecedented rate, and the era of big data has begun. Decision tree is a commonly used data processing model in data mining. Common decision tree construction algorithms include ID3, C4.5, and CART. However, decision tree construction algorithms such as ID3, C4.5, and CART are all for static data sets. By design, they cannot be directly applied to the processing of data streams, which are endless. In addition, the data flow will continuously flow into the system at a very fast speed, which also brings great challenges to the training of decision trees. Existing research also has some solutions to the processing of data streams....

AI Technical Summary

Problems solved by technology

However, since it is impossible to use infinite data sets to calculate the selection of optimal features, only known data can be used to form sample sets to train decision trees, which is far from meeting the needs of current data processing.

Images