Automated Feature Engineering and Deployment with Multiple Datasets | DataRobot AI Accelerators

Automated Feature Engineering is one of DataRobot's most powerful, time-saving features. Using the Python package to interact with the DataRobot API, makes this even more powerful. This demonstration shows how to use multiple data sets for model training and inference. DataRobot will perform cross dataset feature engineering automatically for training and for batch scoring. The link below will take you to the notebook used.

AI Accelerator Link - Access the Notebook being Demonstrated

Speaker
Chandler McCann, Field CTO, DataRobot

Content
0:50 Personal introduction/Chandler McCann
2:10 What are AI Accelerators?
2:45 Examples of DataRobot AI Accelerators
3:31 Where to Find AI Accelerators
4:18 Working with data in many tables
4:42 Orientation to the Use Case
5:25 Data aggregations across time
6:35 Typical machine learning lifecycle
7:20 Feature engineering and deployment challenges
8:30 [MAIN TOPIC] Automated Feature Engineering and Deployment with Multiple Datasets
9:00 Data versioning for the MLOps pipeline
9:45 DataRobot Notebooks
10:28 Versioning data sets pulled from Snowflake
11:34 Repeat for other data sets.
11:58 Time-Aware Feature Engineering
12:50 Feature discovery aggregation settings
13:27 Define feature derivation windows
14:01 Model building
14:50 Newly engineered features
15:24 Examine the top model

Request a Custom Demonstration

Stay connected with DataRobot!