Analytics and Data Summit 2020

The data science process can be highly iterative. Popular methodologies, such as CRISP-DM, have iterative loops built in, especially for data preparation, and model building and evaluation. Fortunately, a significant part of this process can be automated. Such automation not only improves data scientist productivity by addressing some of the more time iterative aspects of data science, but also enables non-experts to leverage machine learning even if they do not know the finer details of the algorithms and their settings. Still, such automation doesn't eliminate the need for competent machine learning skills.

In this session, you will learn about the ML process and what is reasonable to automate today and what still requires the human touch. Learn about the automation provided in Oracle Machine Learning, including automatic data preparation (ADP), automated text processing, automated partitioned models, and automated machine learning (AutoML). The latter focusing on algorithm selection, data feature selection, and ultimate hyperparameter tuning. These capabilities will be highlighted through a series of examples using the SQL, R, and Python APIs. These features all help to reduce the effort, and in some cases, knowledge, required to go from business problem to business solution using machine learning.