Not your parents' machine learning: How to ship an XGBoost churn prediction app in under four weeks
Machine learning is a pivotal technology. However, bringing an ML application to life often requires overcoming bottlenecks not just in the model code but in operationalizing the end-to-end system itself. Goodman Gu shares a case study from a leading SaaS company that quickly and easily built, trained, optimized, and deployed an XGBoost churn prediction ML app at scale with Amazon SageMaker.
| Talk Title | Not your parents' machine learning: How to ship an XGBoost churn prediction app in under four weeks |
| Speakers | Goodman Gu (Cogito) |
| Conference | Strata Data Conference |
| Conf Tag | Big Data Expo |
| Location | San Jose, California |
| Date | March 6-8, 2018 |
| URL | Talk Page |
| Slides | Talk Slides |
| Video | |
Customer churn is very costly to a business. Studies have shown that it typically costs hundreds of dollars to acquire a replacement customer. Early warnings of unhappy customers allows us to incentivize and engage with them to improve satisfaction and retention. Goodman Gu shares a case study from a leading SaaS company that quickly and easily built, trained, optimized, and deployed an XGBoost churn prediction ML app at scale with Amazon SageMaker. Goodman explores the trade-offs in feature engineering, algorithm selection, and hyperparameter tuning, whether it’s better to use a classifier with Keras using a TensorFlow backend or XGBoost, and whether or not to use Spark MLlib. Along the way, Goodman discusses the benefits of Amazon SageMaker, a fully managed end-to-end machine learning service and production pipeline.
