Studies On Theory And Method Of Measurement Model For Online Brand Loyalty

The 13th Five-Year Plan has clearly put forward to insisting the strategy of boosting domestic demand,fully excavating the huge potential of domestic demand,and expanding new-type consumption format.Especially under the background of Internet economy representative "New Economy",online consumption as the representative of new-type consumption format is gradually changing people's consumption concept and mode.Online brand loyalty is extremely important in online consumption domain and of great significance in stimulating consumption,realizing the reform of corporate business models such as precision marketing,risk management and control,decision support,efficiency promotion,and product design,business pattern and even business thinking,and promoting corporate competitiveness in online domain.Application of scientific and effective machine learning method is of urgency and necessary for making systematic analysis and research on online brand loyalty.In big data environment,faced with mass data information provided by online consumption,traditional techniques cannot meet the competitive demand of enterprise creating and maintaining brand loyalty.The following problems are most conspicuous:traditional random sampling method cannot locate consumer group with high brand loyalty.Meanwhile,traditional data processing technique cannot handle online consuming behaviors with volume,mixed and unstructured data.Faced with tremendous online data the problem of how to use real-time online data to position brand consumption group's demands,traditional method has its limitations.This research aims at constructing an AI model in an online consumption big data era--machine learning model.Aggregate online consumers'behaviors of purchasing brand products through machine learning method,and establish online brand loyalty measurement model to realize aggregation of customers with similar brand loyalty and measurement of online brand loyalty.During this process,make a detailed study on machine learning path,machine learning algorithm,model building method and model verification and optimization method.This research focuses on the ideology of theoretical study--model design---model optimization,to mainly complete following research work:(1)On the basis of summarizing domestic and overseas research status of brand loyalty theory,discuss big data features,big data technology and measurement technology of online brand loyalty and big data machine learning method of online brand loyalty.(2)Establishing online brand loyalty measurement model is the subject of this research with processes carried out by following the machine learning path of online brand loyalty measurement.Research processes include:1)Online brand loyalty measurement related online data acquisition.Collect and preserve online brand loyalty measurement related online data through designing spider algorithm.2)Online brand loyalty measurement related online data cleaning.Detect online brand loyalty measurement related online data through designing online data detection algorithm;and conduct online brand loyalty measurement related online data cleaning through designing online data cleaning algorithm.3)Online brand loyalty measurement oriented machine learning method modelling.Based on feature selection basis and characteristic index definition,complete characteristics construction required by the model through designing algorithm;on the basis of giving mathematical definition of brand loyalty measurement model and model aggregation,realize model construction through designing clustering algorithm.4)Brand loyalty measurement model verification.According to the definition of internal validity index and external validity index,realize model interior validity verification through designing internal validity verification algorithm;realize model exterior validity verification through designing external validity verification algorithm.(3)Optimization of online brand loyalty measurement model is the focus of this research.Main research work include:obtaining the optimal model through data processing optimization,feature engineering optimization and algorithm adjustment and optimization,and evaluating the optimal model from two aspects of realizing customer clustering with similar online brand loyalty and realizing the definition of online brand loyalty degree.Innovation points of this research mainly include:(1)Targeting at six particularities of online brand loyalty big data under the "New Economy" with representatives of big data processing,big data value excavation and Internet economy:circumpolar latitude characteristic of online brand loyalty big data and resultant online brand loyalty big data analysis mode,subversive reform produced by analytical thinking and technology,multi-factor's influence generated from online brand loyalty big data,high complexity of online brand loyalty big data analysis model and algorithm.Then make a systematic and in-depth exaction and analysis on these six characteristics for the for the time,so as to lay foundation for further study of online brand consuming behavior and online brand loyalty under the "New Economy" represented by Internet economy.(2)Propose the idea of applying machine learning method in online brand loyalty measurement,and realize the construction of online brand loyalty model through using machine learning method measurement.Improve the model definition on the basis of referring to the concept of multidimensional data cube in data mining domain.It has further expanded the economic application scope of machine learning method.(3)Through full data analysis model,i.e.sample-equal-to-population big data analysis model,realize customer clustering with similar online brand loyalty and the definition of online brand loyalty degree.It has broken through the limitation of traditional random sampling method in positioning high brand loyalty consumption group and the limitation of traditional data processing technology in handling online consuming behaviors with mass mixed and unstructured data characteristics.(4)Brand loyalty measurement model optimization is the focus of this research.During the process,feature engineering optimization is the core of model optimization.In terms of clustering method,there exist problems of feature engineering applicability.This research make adjustment and improvement of detailed characteristic selection method on the basis of referring to the ideology so that it can be better applied for model optimization.The research is innovative.