Prediction model construction method, test method, device and system

A construction method and model technology, applied in the field of prediction model construction, can solve problems such as improper model control or parameters, easy underfitting, easy overfitting, etc., to achieve good model structure, accurate correlation, and avoid underfitting The effect of summing and overfitting problems

Active Publication Date: 2020-08-07
ALIBABA GRP HLDG LTD
View PDF4 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] For linear models such as logistic regression (LR), if there is a relatively large nonlinear relationship in the data law, it is easy to underfit;
[0006] Models such as tree structure and general neural network structure can fit possible nonlinear relationships, but if the model control or parameters are improper, or there are too many parameters, it is easy to overfit

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Prediction model construction method, test method, device and system
  • Prediction model construction method, test method, device and system
  • Prediction model construction method, test method, device and system

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0070] Such as figure 1 As shown, the method of this embodiment may include the following steps:

[0071] Step 110, constructing a neural network model including input nodes and output nodes according to pre-collected sample data.

[0072] Wherein, there are generally multiple sample data collected, and each sample data may include characteristic values ​​and actual values. Specifically, a corresponding number of input nodes may be set according to the number of feature values ​​in each sample data, and one output node may be set.

[0073] Step 120, determining the correlation coefficient between the output value of the non-output node and the theoretical output value of the output node based on the sample data.

[0074] Specifically, the output value of the input node may be determined according to the feature value in the sample data, and the theoretical output value of the output node may be determined according to the actual value in the sample data. When the neural net...

Embodiment 2

[0101] In the above example 1, although the prediction model was obtained through evolution, due to the existence of some accidental factors, for example, the sampling data used is not representative, so it cannot be guaranteed that the obtained model is the optimal model .

[0102] Therefore, in order to further search for a better model, optionally, the process of mutation can be further added, such as image 3 As shown, the following steps may be included:

[0103] Step 144, randomly adding or deleting intermediate nodes and / or inter-node connections in the converged neural network model at least once to obtain a mutated neural network model.

[0104] Among them, the operation of adding intermediate nodes and / or inter-node connections can be referred to above figure 2 The introduction of its related content will not be repeated here. Whereas delete operations that delete intermediate nodes and / or connections between nodes, such as Figure 4 As shown, it can be divided ...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses a prediction model construction method, a prediction model test method, a prediction model test device and a prediction model test system. The prediction model construction method comprises the steps of constructing a neural network model comprising an input node and an output node according to pre-collected sample data; determining a correlation coefficient between an output value of a non-output node and a theoretical output value of the output node based on the sample data; adding an intermediate node and/or inter-node connection in the neural network model based onthe correlation coefficient to obtain an evolved neural network model; and generating the prediction model based on the evolved neural network model. According to the method, a relatively good fittingeffect and a relatively good generalization effect are achieved at the same time by using a minimum model structure.

Description

technical field [0001] The invention relates to a construction method, a testing method, a device and a system for a prediction model. Background technique [0002] In daily business, we often encounter the problem of predicting the characteristic quantity in a certain period of time in the future, such as predicting the road flow in the future, predicting the traffic flow and the flow of people in the airport or port in the future, and predicting the traffic flow in the future The traffic flow and the flow of people migrating between various areas of the inner city, etc. [0003] Taking the forecast of the traffic flow of a certain road in the future as an example, there will be many factors affecting the future traffic flow, for example, it is related to the traffic flow of the current road in the previous period, the traffic before and after the same time period of the previous day, and the adjacent traffic flow. It is related to the current traffic of the road, and it m...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06Q10/04G06Q50/26G06N3/04G06N3/08
CPCG06Q10/04G06Q50/26G06N3/04G06N3/082Y02T10/40
Inventor 贾建超
Owner ALIBABA GRP HLDG LTD
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products