Individual and Group Behavior Based Customer Product Recommendation to Designing Information Systems SPSS Statistics

We introduce an enhanced product embedding representation tailored for retrieval-based product recommendation systems. This study combines method-specific product embeddings into a unified embedding, leveraging text, images, and content such as composite filtering signals to enrich recommendations. By incorporating a fusion step late in our model, independent training of each method is feasible, ensuring a modular architecture conducive to real-world deployment. Recommender systems have become pivotal in navigating data overload, sifting through vast datasets to pinpoint relevant information based on consumer preferences, behaviors, or interactions. They predict a user's affinity for an item based on the user's profile. Utilizing SPSS Statistics, a robust tool for statistical analysis, we evaluate the reliability of our model. Results indicate a Cronbach's Alpha reliability score of 0.450, signifying 45% reliability. Compared to a threshold of 14% derived from literature, our model is acceptable for analysis. In conclusion, the study confirms the efficacy of our model with a Cronbach's Alpha score of 0.450, establishing its potential for real-world application in information systems design.