Predictive Analytics using R with Drupal
summary
Built pipeline uses Google Analytics client IDs merged with CRM records, R (caret/paradox), plumber as an API and Docker/DigitalOcean to host the model, with Drupal front-end calling the API and JavaScript swapping the homepage banner. Models train offline (70/30 split), hit about 60–64% accuracy in this deployment, use a 0.36–0.38 confidence threshold to flip content, and require nightly GA pulls. Early results show flipped-banner clicks convert ~4× better; model files are saved and served as a REST endpoint.
AI-written summary · about
