3.2 Selection of the dynamic factors
To extend the study of the components in the interest chain,we have to analysis the dynamic factors.First of all,many factors can affect the producers.The demand for medical institutions and the cost of production may affect the output of the antibiotics.The government subsidies,the competition of enterprises and the reputation of the enterprise can affect the benefits of the enterprise indirectly.
To the distribution process,the most important factor may be the control from the government.Because of the lax supervision to the sell of the antibiotics,the demands from consumers can account for a large proportion of the profile of medical institutions.
The doctors has two sides in the sell of the antibiotics.On one hand,they want to get more profile;on another hand,they have their own social responsibility.
When it comes to the patients,the consumption level in their medical cure and the doctors advice are the most important factors.The main purpose for farmers to use the antibiotics is to short the culturing period.So,we can see the main dynamic factors that affect the interest chain in the table 2.
Table 2 Dynamic factors
According to the table and the analysis of different parts,we can figure out the factor structure of each part.
Figure 2 Factors for the production enterprise
•The dimensionless processing of data
To analyze the different factors in one model,we have to make the factors dimensionless.There are many ways to unified the dimension.To simplify the calculation process,we use the extremum method.If we use x i to represent the former data,then we get the data after dimension y i：
Among the equation,k and q are the control parameter.
Figure 3 Factors for the medical institution
Figure 4 Factors for the consumer
Except the quantifiable factors,there are many factors which cannot be quantified.We choose the reputation of the enterprise as an example to undertake the scoring systems.When it comes to the reputation of the enterprise,we often take the satisfaction of people as an alternative.Then,we can do a survey to let consumers give their about the enterprise.And the total score is ten.
Δγ means the satisfaction variating range.
And the results can be divided into several levels,which can be seen in the table 3.The quantified dates can also be non-dimensionalized by the method used in equation 1.
Table 3_The evaluation of non-quantification factors
•establishment of the correlation coefficient matrix
To figure out the effect among the factors,we establish the correlation coefficient matrix.Then we can assume the correlation coefficient matrix can be represented as follows.
Through the establishment of the correlation matrix model,we can calculate for each factor with corporate interests between two correlation coefficient matrix.But we still found that some of the relevant risk coefficient changing with corporate interests range is small,even can be neglected.The result of multivariate function relation equation is complex,often do not have a final solution.To eliminate the possibility of no solution,we need to modify the factors that has little impact on the function.As a consequence,we establish the expert-principal component model.The linear correlation matrix can be calculated by the linear coefficient model.
In order to calculate the characteristic equation,let：
The eigenvalue can be represented asη.The corresponding eigenvector is|Ei|=1.After getting the eigenvalue,we can calculate the contribution rate and cumulative contribution rate.When the contribution rate and cumulative contribution rate are range from 85%～95%.The corresponding factor can be considered as a proper result.And the contribution rate can be represented as：
The cumulative contribution rate can be represented as：
By calculating the contribution rate and cumulative contribution rate,we can confirm the principal component.Other factors except the principal components can be ignored.And the principal component can be represented as matrix：
