Product competitiveness analysis for e-commerce platform of special agricultural products

On the basis of analyzing the influence factors of the product competitiveness of the e-commerce platform of the special agricultural products and the characteristics of the analytical methods for the competitiveness of the special agricultural products, the price, the sales volume, the postage included service, the store reputation, the popularity, etc. were selected in this paper as the dimensionality for analyzing the competitiveness of the agricultural products, and the principal component factor analysis was taken as the competitiveness analysis method. Specifically, the web crawler was adopted to capture the information of various special agricultural products in the e-commerce platform ---- chi.taobao.com. Then, the original data captured thereby were preprocessed and MYSQL database was adopted to establish the information library for the special agricultural products. Then, the principal component factor analysis method was adopted to establish the analysis model for the competitiveness of the special agricultural products, and SPSS was adopted in the principal component factor analysis process to obtain the competitiveness evaluation factor system (support degree factor, price factor, service factor and evaluation factor) of the special agricultural products. Then, the linear regression method was adopted to establish the competitiveness index equation of the special agricultural products for estimating the competitiveness of the special agricultural products.


Introduction
In recent several years, in order to further stimulate domestic economic development, the government proposed the slogan of "Internet +", "Mass Innovation" and "Mass Entrepreneurship". In fact, the mode of "Internet + Traditional Agriculture" has become an important method for converting present agricultural development mode and promoting the efficient allocation of market resources. Under such situation, the e-commerce of agricultural products has accordingly become a new growth point of the "Agricultural Economy (related to agriculture, farmer and rural area)" in China. At present, the two major e-commerce leaders in China -----Jingdong ("Fresh and Specialty Sectors") and Taobao ("chi.taobao.com") are energetically promoting the agricultural products on e-commerce platform; meanwhile, some particular e-commerce platforms for agricultural products, such as www.womai.com, www.tootoo.cn, www.sfbest.com, www.benlai.com, www.taocz.com, www.fieldschina.com, Long Bao Tracing Mall and www.cndlbz.com, are widely established. In order to further promote the development of the e-commerce platforms particularly for special agricultural products, this paper is mainly focused on researching the competitiveness analysis system on the e-commerce platform of the special agricultural products in order to provide relevant services to the e-commerce platform of the special agricultural products.

Introduction to principal component factor analysis method
In practical problem research, problem variables may have certain correlation, namely: the information described by the variables may be overlapped with each other [17] . In order to solve variable correlation problem, it is necessary to adopt less variables to replace original variables for variable conversion under the precondition of not losing or only losing very few original variable information. Therefore, the factor analysis method is adopted to solve such problem.
The factor analysis method is a dimension reduction method for the problem describing variables. In this method, the internal correlation of the variables is mathematically analyzed and only several abstract variables (interpretable) are adopted to describe all or main information of original variables, wherein such interpretable abstract variables are called as factors in the factor analysis method. Original variables can be directly obtained from the observed values of the experiment samples, and the abstract variables are interpretable in practical problems.
In the factor analysis method, the most universal factor extraction method is the principal factor analysis method. Specifically, the original variables of the samples are standardized to convert the original variable group into one group of irrelevant factor variables; then, the principal factors are selected according to the completion degree of the factor group for describing the variable information to reduce the dimension of the original variables while not losing the information of the original variables.

Characteristics of principal component factor analysis method
(1) Original variables are correlated to each other, but the factor variables are not significantly correlated to each other, thus favorable for deeply analyzing the problem; (2) The number of the factor variables must be less than that of the original variables in order to simplify problem analysis.
(3) The factor variables must be able to describe most information of the original variables in order to maximally reduce original information loss and improve problem analysis accuracy.
(4) The factor variables must be interpretable, namely: the factor variables can describe a certain aspect of the practical problem.

Model analysis
p original variables, x 1 x 2 … x p (namely, p influence factors of the observed data of each special agricultural product), are set in this paper, and each original variable includes m aspects of the commodity competitiveness, recorded as F 1 F 2 … F m (m < ), wherein m aspects are called as the common factors of p original variables, and the aspects that cannot be interpreted by m common factors are called as the special factors of the original variables, recorded as ε 1 ε 2 … ε p . Therefore, the factor analysis model is established as follows: Where a ij (i = 1,2 … , m; j = 1,2 … , p) is the factor load, called as the load of the i th variable on the j th common factor, namely, the degree of the i th variable for interpreting the j th aspect of the commodity competitiveness.
The matrix expression of this model is as follows: Description: The following conditions should be met: (1) ≤ , namely: the number of the common factors must be less than that of the influence factors in original data; (2) ( , ) = 0, namely: the correlation between the common factors and the special factors must be 0;  , namely: all special factors are not correlated to each other, but the variance may not be the same.

Process analysis
The product competitiveness analysis process of the e-commerce platform of the special agricultural products is as shown in Fig Agricultural Products (1) Standardize original data to obtain standard matrix X: various data in original variables are not uniformly measured, so it is necessary to standardize the original data to convert them into standard matrix in order to reduce analysis difficulty and implement uniform data calculation; (2) Calculate correlation coefficient matrix R according to the standard matrix: set |R-λE|=0, calculate the characteristic value, the contribution rate and the accumulative contribution rate of R, and determine the number of the common factors according to the principle that the characteristic value is not less than 1 or the accumulative variance contribution rate is not less than 85%.
(3) Calculate the characteristic vector and initial factor load matrix A; (4) Estimate factor scores by regression method, and take the specific value of the variance contribution rate of each factor and the total variance contribution rate of the factors as the weight value for the weighted summary in order to obtain the factor analysis model; (5) In case of unobvious factor meanings, adopt the maximum variance method for the orthogonal rotation of the initial factors in order to obtain rotated common factor solution B; (6) Comprehensively analyze and evaluate the factors according to the sample factor scores.

Empirical analysis
The factor analysis function provided by SPSS tool was adopted and 6 data indexes ----commodity price, commodity sales volume, user evaluation, postage included service, store reputation and commodity collection quantity were taken as the original variable. Then, the principal component factor analysis method was adopted to convert the original variables and calculate the factor load matrix. The common factors obtained at the first time were not obviously interpretable, so the maximum variance method was adopted to rotate the load matrix. Finally, the regression method was adopted to calculate the influence factor scores.
According to the results of KMO and Bartlett sphericity tests (see Tab.4.1), KMO value is more than 0.6, thus indicating high variable intercommunity. Meanwhile, the significance level (0.00) of Bartlett sphericity test is less than 0.05 and the hypothesis of Bartlett sphericity test is refused, so the original data of the special agricultural product samples collected thereby are applicable to factor analysis.
Tab According to the factor load matrix of the experiment result (see Tab.4.4), the interpretations of the four common factors in the variables are dispersed and have unobvious factor expressions, so the factors cannot be directly used for interpretation. Subsequently, the maximum variance method was adopted for the orthogonal rotation of the initial factors, without changing the accumulative contribution rate of the rotated principal components (see Tab.4.3). Therefore, the sample data information of the special agricultural products will not be further lost.
Tab The result of the initial factors after orthogonal rotation is displayed in the rotated factor load matrix table (see Tab.4.5). According to the table, the rotated load is obviously polarized towards 0 and 1. In the rotated matrix table, the factor attribution of each variable can be easily judged. Factor 1 mainly has large load for commodity sales volume, store level and commodity collection quantity, and these variables mainly represent the popularity degree of the product and the support degree of the buyers in e-commerce platform, so F1 can be interpreted as the support degree factor. Factor 2 mainly has large load for commodity price, so F2 can be interpreted as price factor. Factor 3 mainly has large load for transportation expense, so F3 can be interpreted as service factor. Factor 4 mainly has large load for good reputation rate, so F4 can be interpreted as evaluation factor.
According to the experimental analysis result, the first common factor is the support degree factor, orderly followed by the price factor, the service factor and the evaluation factor. The degree of the influence of the factors on the competitiveness of the agricultural products is as follows: support degree > price > service > evaluation. The rotation component matrix can reflect the proportion of each principal factor, and we can obtain the following principal factor model for the product competitiveness: (1) The original information of the special agricultural products mainly includes commodity price, commodity sales volume, user evaluation, postage included service, store reputation and commodity collection quantity. In this paper, the principal component factor analysis method was adopted to select four principal factors (support degree factor, price factor, service factor and evaluation factor) to describe the competitiveness of the special agricultural products. In this way, only a few abstract variables were adopted to describe all information in order to summarize the main influence factors of the competitiveness of the agricultural products.
Tab Common factors (F1, F2, F3 and F4) are variables, but they are different from such original variables as commodity sales volume and commodity price and cannot be directly used as statistical information. However, the factor scores, namely the score of each sample on each common factor, can be calculated to measure the competitiveness of the commodity on the support degree factor, the price factor, the service factor and the evaluation factor. According to the component score coefficient matrix (see Tab.4.6) obtained by regression method, the regression equation of the common factor scores can be obtained as follows: The factor scores can be obtained according to the regression equation of the principal factors, and the weighted mean of the contribution rate of the common factors to original information is taken as the weight value to calculate the comprehensive score of the product competitiveness of individual sample data for the comprehensive evaluation of the competitiveness of the agricultural products.
The competitiveness of the special agricultural products is jointly influenced by four factors. In order to objectively measure the product competitiveness, the specific value of the variance contribution rate of each factor and the total variance contribution rate of the factors is taken as the weight value to calculate the comprehensive score as the product competitiveness index.

Conclusion
In this paper, the principal component factor analysis method was adopted to establish the product competitiveness analysis model for the e-commerce platform of the special agricultural products, and SPSS statistical software was adopted as the analysis tool to research and analyze product competitiveness. According to the research and analysis result, the competitiveness of the special agricultural products is mainly influenced by four factors, namely support degree factor, price factor, service factor and evaluation factor. Different products are differently influenced by the four factors, thus presenting the competitiveness difference of the special agricultural products on different factors. Meanwhile, the comprehensive evaluation model for the competitiveness of the special agricultural products was also established in this paper to analyze the competitiveness index of the products in the e-commerce platform of the special agricultural products so as to measure the competitiveness of the special agricultural products.