Learn to Use Databricks for the Full ML Lifecycle
0 likes880 views
The document discusses utilizing Databricks for managing the full machine learning lifecycle, emphasizing essential capabilities such as data processing, secure collaboration, and documentation. It presents a business context involving customer retention analysis and the need to predict customer churn using historical data. The document also mentions upcoming blogs on MLOps and machine learning engineering topics.
Learn to Use Databricks for the Full ML Lifecycle
- 1. Learn to Use Databricks for the full ML Lifecycle Rafi Kurlansik, Sr. Solutions Architect
- 2. Structured Semi-structured Unstructured Streaming BI & SQL Analytics Machine Learning Real-time Data Applications Data Management & Governance Open Data Storage Data Science & Engineering Lakehouse Platform Simple | Open | Collaborative Reliable | Scalable | Secure
- 3. Structured Semi-structured Unstructured Streaming BI & SQL Analytics Machine Learning Real-time Data Applications Data Management & Governance Open Data Storage Data Science & Engineering Lakehouse Platform Simple | Open | Collaborative Reliable | Scalable | Secure Our focus today
- 4. Essential Capabilities for Full ML Lifecycle To lower risk and maintain stability of our ML pipeline, we need to think about: ● Robust Data Processing and Management ● Secure Collaboration ● Testing ● Monitoring ● Reproducibility ● Documentation ...for code, data and models.
- 5. Business Context: Customer Retention You are on a marketing analytics team and you have a lot of demographic and historical service data on your customers that have churned, which has been put into a SQL Analytics dashboard. The data team has been asked by business stakeholders if you can go further and predict which customers are likely to churn. Knowing this will allow the business to take action and retain revenue. Sounds simple enough. What steps do we need to take?
- 11. To the demo we go!
- 12. The Full ML Lifecycle
- 13. To learn more: We’ll be releasing a series of blogs on MLOps and ML Engineering throughout 2021: ● The Need for Data-centric ML Platforms ● Selecting Technologies and Platforms for Data Science and Machine Learning ● Model and Data Monitoring on Databricks ● … and more Check out the other MLOps talks at DAIS!
- 14. Thank you!