Research On Brand Status Discrimination Model Based On FCM Clustering

The top priority of corporate brand sustainable development strategy is realize brand market status in real time.At present,with the sudden increase of corporate sales data and the increase in the complexity of brand market influence factors,a single subjective judgment of corporate decision makers has been unable to accurately and timely determine the status of the corporate brand market.Therefore,this paper proposes a framework for the collection and brand market analysis system based on the massive sales data of the corporate brand market.This article has completed the collection,storage,and processing of sales data for the brand market.And this article has extracted decision factors that affect brand market status.Taking advantage of the brand's life cycle,brand market position,and the brand's price as the dimension,an optimal discriminative model of the brand market state is established that provide comprehensive and accurate decision-making basis for corporate brand development.The main research content is as follows:1)Data Acquisition and Processing: More than 200 million sales data collected each month for the national brand market,including sales data for retail terminal data and consumer environments.And perform preprocessing of massive data sources that solved the problem of normalization of data and data storage.And analyzed the data mining process of the brand market state.2)Determine brand market status decision-making decision factor: In order to reduce the complexity of the factors affecting the brand's market state,this paper uses Principal Component Analysis(PCA)to reduce the dimensionality of brand status attributes and determine the nine characteristics that affect the brand market status of the company.3)Analysis of brand market status: Using FCM fuzzy clustering algorithm for multi-dimensional clustering analysis of brand market status.Determine metrics for brand market status.Combining enterprise's integrated marketing system to construct CM discriminant model of brand market status.Position the brand's market status accurately,formulating brand marketing strategies,ensuring the stable development of corporate brands,and enhancing the competitiveness of corporate brands.4)Verify the scientificity and reliability of the brand status discriminant model: Achieve parallel analysis of sales data in the brand market on the Spark platform.And using four major evaluation indicators to evaluate the reliability of the discriminant model which was selected with good stability,high extensibility,and high accuracy.5)The improvement of the algorithm,including the FCM clustering algorithm and the NB classification algorithm: The method of adjusting the sample density distribution value is adopted to improve the unbalanced sample base defects existing in the FCM algorithm and improve the accuracy of the algorithm clustering results.The dichotomy method is used to improve the disadvantage that the larger the amount of data in the NB classification algorithm is,the worse the classification performance is.It mainly simplifies the probabilistic condition of class conditions,reduces the computational complexity of the algorithm,and improves the efficiency of algorithm classification.6)Designed and implemented an enterprise brand marketing integration platform: The brand status discriminant model was applied to the platform,and all-round and in-depth analysis and display of the brand was carried out to achieve a precise positioning of the company's brand market and increase the sales profits of the company.