| With the development of market economy and the progress of science and technology,customers,as an important asset,begin to receive more and more attention from enterprises.Customer assets have become one of the key factors for enterprises to evaluate their own development status,adjust the development direction and promote their sustainable development.Therefore,for enterprises,in-depth analysis of customer behavior is not only conducive to enhancing the competitiveness of enterprises in the current market,but also can help reduce the risks existing in the unpredictable market environment.At present,the focus of customer behavior analysis is mainly focused on the classification and value analysis of customers through the use of historical data.The analysis method is relatively simple,lacks flexibility,and the analysis results are relatively rough.At the same time we should pay attention to,as people material life rich people demands of personalized,customers' preferences from single to diversified,their shopping features and differences in behaviour is becoming more and more big,in this context,the single customer behavior analysis has been unable to adapt to changing market demand at present,and then improve the previous analysis way to adapt to the diversification of market demand is what we want attention and research.In addition,with the increasing demand for customer behavior analysis and the deepening of marketing research,consumer involvement has been more and more mentioned in recent years,and its importance has been highlighted.From consumers,the higher the degree of involvement with the product,in searching for information and the more you will participate in the decision to leave "trace",the "trace" may be left in your search engine search frequency,were also more likely to interact on social platforms,these will definitely leave "trace" helps enterprises to make more accurate sales forecast,so a high involvement product sales forecast to scholars more and more attention.Mainstream research has shifted from focusing on involvement and consumer behavior to involvement and sales forecasting.Researchers are also constantly introducing the information resources brought by the era of big data and modern technological means to conduct sales forecasting research.We have noticed that when conducting customer behavior analysis and sales prediction,scholars often only pay attention to the single influence of customer behavior analysis on sales prediction,and basically ignore the dialectical relationship between the two.In fact,it is helpful for enterprises to make more efficient prediction to clarify the dynamic relationship between the two.In this paper,customer behavior analysis and enterprise sales prediction are respectively summarized.Based on the results of literature review,the research questions of this paper are put forward.The existing prediction models are introduced in detail and the models and methods suitable for this study are selected through comparison.On the one hand,based on the RFM analysis model,a total of 9 singlephase customer purchasing behaviors of 9 states are divided according to the two behavioral variables F and M.On the other hand,k-means algorithm is adopted for customer clustering analysis,and five schemes are further summarized for different customer states.In addition,in this paper,based on the markov chain theory is needed to establish transfer matrix,the transition probability estimates based on the Dirichlet-Multinomial model,and estimate the parameters of the model text using gibbs sampling algorithm.Finally,through modeling analysis and parameter optimization.On the other hand,based on the state transition probability matrix of markov prediction model,the weighted sum of each item is carried out through the weighted thought,and the weight is optimized by the genetic algorithm theory to achieve the expected sales prediction results.Finally,the dynamic dialectical relationship between them is analyzed through the empirical results. |
PDF Full Text Request