Developing a model is only a small part of successful Data science project. Successful data science project also includes efficient data preparation and exploration, extensive modeling and tuning, controlled production deployment, extensive monitoring and model retraining, as well as governance of the entire process. In this talk we’ll look into productionization of data science products end to end and how it can be done efficiently using Databricks. We’ll cover the topics of data preparation with Spark and Koalas, experiment and model tracking with MLFlow, representation of results with DBSQL dashboard. We’ll also take a look at advanced features of Databricks like Auto ML.
Supported by
