Table of contents
Gray relational analysis is used for system analysis
Gray relational analysis is used for comprehensive evaluation
Try not to use gray correlation analysis in the US competition, because it is proposed domestically and the country may not know about it.
Gray relational analysis is used for system analysis
Purpose: Get which independent variable (subsequence) has the greatest impact on the dependent variable (parent sequence)
System analysis methods: regression analysis, principal component analysis, variance analysis
The above methods have shortcomings:
they require a large number of samples and the samples obey a certain typical probability distribution; they require a large amount of calculation; the quantitative results may not match the qualitative analysis .
The basic idea of gray correlation:
judge whether the connection is close based on the similarity of the curve shape. The closer the curve is, the greater the correlation.
When to use standardized regression/gray correlation:
example
step
1. Use excel to draw a line chart of the original data + simple analysis:
2. Determine the analysis sequence
- Parent sequence/reference sequence/parent indicator: equivalent to dependent variable
- Subsequence/reference sequence/subindicator: equivalent to independent variable
3. Preprocess data (including parent-child sequences)
- Purpose of preprocessing: remove dimensional effects, reduce variable calculation range and simplify calculations
- Preprocessing method: first find the mean of each indicator, and then divide each element in the indicator by its mean
4. Calculate the correlation coefficient and gray correlation degree between each indicator in the sub-sequence and the parent sequence
- Minimum difference between two levels, maximum difference between two levels
- Correlation coefficient
- Gray correlation degree: The largest gray correlation degree is the independent variable with the greatest influence.
Gray relational analysis is used for comprehensive evaluation
When there is no data: AHP analytic hierarchy process
When there is data: TOPSIS, gray correlation
example
step