ControlExpert (CE) is a service provider that aims for an entire digitalization of the claim process in automotive area. As a partner of insurers CE receives claim documents, digitalize and verifies them. Target of this paper is to give first-hand insights of a predictive model development from early stage in a lab to final deployment into production environment at CE. From understanding business and data context over model creation and improvement up to deployment as a web service typical problems are tackled and solutions are outlined. Finally paper describes key success factors of the project.
Required audience experience: No specific knowledge required.
Objective of the talk: Delegates will get first-hand insights from a real machine learning project. The talk outlines necessary steps from first experiments in a lab to final production environment of a business process.
Keywords: Predictive Analytics, Supervised Learning, practical machine learning, CRISP-DM, parallel computing, R, R Studio, model performance metrics, ROC Curve, R caret ML package, feature engineering, model error analysis, R web service rApache, deployment production system, key success factors
You can view Stephen’s slides via the link below:
Stephen Seiler: Practical Machine Learning from Lab to Production
You can view Stephen’s presentation below: