Analytics and Data Summit 2019

Oracle has been embedding machine learning algorithms into the SQL kernel of the Oracle Database for many years. Oracle Advanced Analytics 18c delivers significant performance gains and new algorithms including Random Forests, Neural Networks, Unsupervised Feature Selection, Partitioned models and Market Basket Analysis improvements. Data scientists can build models in-database on billions of records in minutes and “scores” millions of records in seconds. Now, Oracle Machine Learning, bundled in the Oracle Autonomous Database, adds collaborative machine learning notebooks for data scientists. Oracle also has added a Python API (OML4Py), ORAAH and new Cognitive Analytics for images, text and deploying models as Microservices. Oracle “moves the algorithms to the data” for hybrid data management and machine learning platforms that enable enterprise-wide machine learning model deployment. Come hear about the latest developments and what’s coming next from Oracle.

The Taxi Fare prediction problem is a well known in machine learning (ML): we have to design, train and test a ML model to predict the taxi fare a customer of the NY City cab company will pay for a ride. This is a basic example used in several ML introductory courses and also ML and artificial intelligence (AI) websites as foundation for more advanced topics. In this presentation, we will work with the publicly available data to have it prepared to use the ML functions of Oracle Advanced Analytics embedded in Oracle Database and, more importantly, the new neural network algorithm introduced in 18c. We will see how the data was prepared, some powerful tools included in the Oracle Data Miner component of SQL*Developer, the PL/SQL code involved to create the model and, of course, some testing. All with a live demo.

Learn the fast and easy way to get started performing machine learning with Oracle SQL Developer 18.3’s Oracle Data Miner drag-and-drop workflow UI. In this introductory hands-on lab, learn from experts who will provide one-on-one coaching, guidance, and instruction as requested in several easy to follow Hands-On-Lab Tutorials. Learn how to explore data and apply machine learning algorithms for common ML use cases e.g. prediction, clustering, key factors identification, anomaly detection and market basket analysis. Follow step-by-step tutorial instructions in a machine learning-friendly environment. Ask questions, get answers and come away with basic understanding and a comfort level for using Oracle’s machine learning capabilities on your own data and use cases.

Try out the new Oracle Machine Learning Zeppelin-based notebooks that come with the Oracle Autonomous Database in this introductory Hands-On-Lab. Oracle Machine Learning extends Oracle’s offerings in the cloud with its collaborative notebook environment that helps data scientist teams build, share, document, and automate data analysis methodologies that run 100% in the Oracle Autonomous Database. Interactively work with your data, build, evaluate and apply machine learning models. Import, export, edit, run and share Oracle Machine Learning notebooks with other data scientists and colleagues all on the Oracle Autonomous Database. Share and further explore your insights and predictions using the Oracle Analytics Cloud.

Oracle develops a number of business applications for organizations that deliver asset maintenance and/or repair. These applications provide planning, scheduling, optimization and controls that can save large organizations many millions of dollars, extend the lifespan of assets and ensure assets work when they are needed. Many maintenance and repair business cases are particularly well suited to using artificial intelligence (AI) and machine learning (ML) tools: optimizing maintenance schedules requires predicting when asset classes and parts will fail; predicting individual asset failures with enough time to prevent downtime requires identifying patterns and anomalies in the IoT sensor data stream; providing the right first-time fix requires predicting the root cause of failure and how best to fix it based on symptoms and circumstances; and many more examples. This session will explore Oracle's various applications, business cases and AI and ML tools for maintenance and repair.

Most machine learning projects fail to go beyond the data scientist. They hit a wall when they try to “operationalize” the ML models into enterprise production IT environments. Open source Python and R are great for data scientists but create problems with the DBA. This dichotomy between IT and data science is eliminated when the ML algorithms are part of the enterprise data management platform. By “moving algorithms to the data”, Oracle has extended the Oracle Database, Autonomous Database and Big Data platforms into hybrid data management and machine learning platforms. Data scientists and application developers can use their language or UI of choice (R, Python, SQL, “drag and drop” UIs and notebooks) to cooperatively build, develop and deploy ML models on premise or in the cloud. Learn how data scientists move beyond “ML tools” and now collaborate with IT, DBAs and application developers to deliver ML enabled applications. Several “Predictive Applications” will be shown.

Oracle Autonomous Databases automate everything, right? So what’s a DBA to do now? DBAs spend the majority of their time on maintenance, security and outage issues leaving little time for innovation. Autonomous Databases are self-driving, self-securing and self-repairing. Autonomous Databases free database developers to leverage valuable data skills to extract more insights and make predictions. Oracle Machine Learning Zeppelin notebooks supports the six basic machine learning steps: business understanding, data understanding, data preparation, model building, model evaluation and model deployment. Database developers perform these tasks now, actually 80% of the work, but refer to them as ETL, data wrangling and production scripts. Come learn how Oracle Machine Learning enables you to transition from database developer to “data scientist” in 6 weeks.