This paper addresses the problem of predicting consumer behavior in live webcast banding and explores the application of machine learning algorithms in this field and their effectiveness. The study mainly focuses on the use of two machine learning models, Decision Tree and Random Forest, to analyze and predict consumer purchase behavior in live streaming bandwagon campaigns. By simulating a dataset containing 15,000 records, this study conducted detailed training and testing of the models to evaluate their performance on key metrics such as accuracy, recall, precision and F1 score. The experimental results show that the Random Forest model outperforms the Decision Tree model on all the evaluated metrics, especially on accuracy and F1 score. This finding highlights the advantages of integrated learning methods when dealing with complex datasets. In addition, the feature selection process revealed that live streaming viewing duration, number of interactions, and anchor ratings are key factors influencing consumers' purchasing decisions. The findings of this study provide valuable insights for the optimization of marketing strategies for webcast bandwagons. By understanding the key factors influencing consumer purchasing behavior, e-commerce platforms and companies can more effectively position their target markets and optimize their advertising and product promotion strategies.

With the rapid development of IoT technology, the e-commerce sector has encountered unprecedented opportunities and challenges. This study focuses on Xiaomi smart bands, collecting health and behavior data from 50 users over a three-month period. By integrating decision tree classification, K-means clustering, and association rule mining, we constructed a systematic brand-building analysis model. The results show that the decision tree classification model achieved an accuracy of 87.5%, effectively identifying user groups with high brand interaction frequency. K-means clustering further segmented these high-interaction users into three distinct groups, revealing significant differences in user behavior characteristics. Additionally, association rule mining uncovered key behavioral patterns, such as users with more than 8,000 steps per day being more likely to purchase sports shoes (support: 0.3, confidence: 0.85). These findings indicate that the proposed model can provide strong data support for brands in developing personalized marketing strategies and enhancing user experience. Although there are some limitations in data sources and model generalization, this study demonstrates the significant potential of combining IoT and data mining technologies in advancing e-commerce. Future research will aim to expand the dataset and optimize the model to enhance its applicability and practicality.

As a fresh engine for economic growth, the digital economy is playing an important role in pushing forward common prosperity. Using provincial panel data of China covering 2011 to 2023, this paper adopts the entropy method, the coupling coordination degree model and a recurrent neural network model to assess and predict how the digital economy and common prosperity move together. The research shows, first of all, that although obvious regional gaps exist, both indicators keep climbing. The eastern part stays in a leading position, while the north-west still meets many difficulties in development. Second, the coupling coordination analysis tells us the two systems used to be out of balance but have been getting better step by step from 2011 to 2023. At present the relation is in a shift period and can be easily influenced by inner factors such as weak digital infrastructure. Last but not least, the RNN forecast suggests that the main coordination index will keep on rising. Therefore, it is of great necessity to speed up digital infrastructure construction and perfect the income distribution mechanism so as to realize higher-quality joint progress.