Data Processing Method and Computer System

Patent metadata

Field	Value
Publication number	US-2016012352-A1
Application number	US-201514852841-A
Country	US
Kind code	A1
Filing date	Sep 14, 2015
Priority date	Jun 27, 2014
Publication date	Jan 14, 2016
Grant date	—

Abstract

A data processing method and a computer system. The computer system may perform discretization processing on a data sample to obtain a data sample in a matrix form, train the data sample in the matrix form according to a preset classification method to obtain a classification rule set, and after converting the classification rule set into a classification rule set that can be recognized by a data decision-making platform, provide the classification rule set to the data decision-making platform, so that the data decision-making platform can perform data decision-making according to the classification rule set that is obtained by the computer system by conversion and can be recognized by the data decision-making platform. All the foregoing processes are automatically completed by the computer system, which avoids human participation.

First claim

What is claimed is:

1 . A data processing method, comprising:
performing, by a computer system, discretization processing on a first data sample to obtain a second data sample in a matrix form; training, by the computer system, the second data sample in the matrix form according to a preset classification method to obtain a first classification rule set; converting, by the computer system using an expression form that can be recognized by a data decision-making platform, the first classification rule set into a second classification rule set that can be recognized by the data decision-making platform; and providing, by the computer system, the second classification rule set that is obtained by conversion and can be recognized by the data decision-making platform to the data decision-making platform.

2 . The method according to claim 1 , wherein the preset classification method is a decision tree algorithm, wherein training, by the computer system, the second data sample in the matrix form according to the preset classification method to obtain the first classification rule set, and converting, by the computer system using the expression form that can be recognized by the data decision-making platform, the first classification rule set into the second classification rule set that can be recognized by the data decision-making platform comprise:
training, by the computer system, the second data sample in the matrix form according to the decision tree algorithm to obtain the first classification rule set in a decision tree form; and converting, by the computer system using the expression form that can be recognized by the data decision-making platform, the first classification rule set in the decision tree form into the second classification rule set that can be recognized by the data decision-making platform.

3 . The method according to claim 2 , wherein the computer system comprises:
a master computing node; and a plurality of decision tree computing nodes, wherein training, by the computer system, the second data sample in the matrix form according to the decision tree algorithm to obtain the first classification rule set in the decision tree form comprises: sending, by the master computing node, a decision tree computing command to each of the decision tree computing nodes; and training, by each of the decision tree computing nodes according to the decision tree computing command and using the decision tree algorithm, a part of the second data sample in the matrix form to obtain the first classification rule in the decision tree form, and wherein the first classification rule set in the decision tree form is a set of the first classification rules in the decision tree form obtained by all of the decision tree computing nodes.

4 . The method according to claim 3 , wherein sending, by the master computing node, the decision tree computing command to each of the decision tree computing nodes, and training, by each of the decision tree computing nodes according to the decision tree computing command and using the decision tree algorithm, the part of the second data sample in the matrix form to obtain the first classification rule in the decision tree form comprise:
acquiring, by the master computing node, an algorithm configuration parameter, wherein the algorithm configuration parameter comprises information about training samples in the second data sample in the matrix form and information about an attribute that participates in decision tree generation; sending, by the master computing node, the decision tree computing command to each of the decision tree computing nodes, wherein the decision tree computing command carries the algorithm configuration parameter; determining, by each of the decision tree computing nodes according to the algorithm configuration parameter carried by the decision tree computing command, the training samples and the attribute that participates in decision tree generation from the second data sample in the matrix form; and training the training samples determined by each of the decision tree computing nodes according to the attribute that participates in the decision tree generation, to obtain the first classification rule in the decision tree form.

5 . The method according to claim 4 , wherein the method further comprises:
sending, by the master computing node, a test command to each of the decision tree computing nodes after it is determined that each of the decision tree computing nodes obtains the first classification rule in the decision tree form; obtaining, by each of the decision tree computing nodes, a test sample set from the second data sample in the matrix form according to the test command; testing the test sample set using the first classification rule in the decision tree form obtained by each of the decision tree computing nodes, to obtain a test result set; acquiring, by the master computing node, the test result set obtained by each of the decision tree computing nodes; determining, by the master computing node, test accuracy according to a preset voting rule and the test result set obtained by each of the decision tree computing nodes; and converting, by the master computing node, the first classification rule set into the second classification rule set that can be recognized by the data decision-making platform when the test accuracy is within a preset proper range.

6 . The method according to claim 5 , wherein the information about the training samples in the second data sample in the matrix form comprises:
a storage address of the second data sample in the matrix form; a ratio of the training samples to test samples in the second data sample in the matrix form; and a ratio of a randomly acquired sample to the second data sample in the matrix form.

7 . The method according to claim 3 , wherein converting, by the computer system using the expression form that can be recognized by the data decision-making platform, the first classification rule set in the decision tree form into the second classification rule set that can be recognized by the data decision-making platform comprises converting, by each of the decision tree computing nodes according to an instruction of the master computing node or a preset conversion policy and using the expression form that can be recognized by the data decision-making platform, the first classification rule in the decision tree form obtained by each of the decision tree computing nodes into the second classification rule that can be recognized by the data decision-making platform, and wherein the second classification rule set that can be recognized by the data decision-making platform is a set of the second classification rules that are obtained by all of the decision tree computing nodes and can be recognized by the data decision-making platform.

8 . The method according to claim 7 , wherein providing, by the computer system, the second classification rule set that can be recognized by the data decision-making platform to the data decision-making platform comprises:
acquiring, by the master computing node, the second classification rule that is obtained by each of the decision tree computing nodes and can be recognized by the data decision-making platform, to obtain the second classification rule set that can be recognized by the data decision-making platform; and providing, by the master computing node, the second classification rule set that can be recognized by the data decision-making platform to the data decision-making platform.

9 . The method according to claim 7 , wherein providing, by the computer system, the second classification rule set that can be recognized by the data decision-making platform to the d

Assignees

• Huawei Tech Co Ltd

Inventors

• Peng Yuan
• Xie Yuanzhi

Classifications

• G06N5/045

  Explanation of inference; Explainable artificial intelligence [XAI]; Interpretable artificial intelligence · CPC title

• G06N99/005 Primary

  Physics · mapped topic

• G06N20/20

  Ensemble learning · CPC title

• G06N5/025 Primary

  Extracting rules from data · CPC title

• G06N20/00 Primary

  Machine learning · CPC title