Analysis of Social Media Consumer Purchase Intention Based on SEM+Machine Learning

In the field of social media marketing, accurately predicting and analyzing consumers’ purchasing intention is very important for companies to formulate effective marketing strategies. This study introduces a social media consumer purchasing intention analysis method based on the combination of structural equation model (SEM) and random forest (RF). This method more accurately predicts and analyzes consumers’ purchasing intention by integrating the advantages of SEM in understanding complex relationships with the ability of RF in processing big data and identifying nonlinear relationships. This study first collects data on consumers’ reactions to social media ads and their purchase intentions through social media platforms. Next, the data is analyzed using SEM to quantify the degree to which variables such as perceived value and consumer co-creation affect purchase intentions and reveal the inherent connections between them. At the same time, potential variables and mediating effects that may exist in the model are identified. Then, the RF algorithm is optimized based on the analysis results of SEM, allowing the RF algorithm to learn from the data and identify the key features and patterns that affect purchase intention. Finally, SEM and RF are combined to build a model that can accurately predict and analyze the purchase intention of social media consumers. The SEM+RF model performs better than the single SVM model in the analysis of social media consumer purchase intention, with higher analysis accuracy and shorter analysis time. Through the variable influence weight analysis, this paper finds that consumer co-creation behavior has the greatest impact on purchase intention (28%), followed by advertising content quality (22%), while advertising exposure (20%), perceived value (18%) and online word of mouth (12%) also have an impact on purchase intention.