Oracle is adding new machine learning features to its data analytics cloud service MySQL HeatWave.
MySQL HeatWave combines OLAP (online analytical processing), OLTP (online transaction processing), machine learning, and AI-driven automation in a single MySQL database.
The new machine learning capabilities will be added to the service’s AutoML and MySQL Autopilot components, the company said when it announced the update on Thursday.
While AutoML allows developers and data analysts to build, train and deploy machine learning models within MySQL HeatWave without moving to a separate service for machine learning, MySQL Autopilot provides machine learning-based automation to HeatWave and OLTP such as auto provisioning, auto encoding, auto query plan, auto shape prediction and auto data placement, among other features.
AutoML augments time series forecasting via machine learning
The new machine learning-based capabilities added to AutoML include multivariate time series forecasting, unsupervised anomaly detection, and recommender systems, Oracle said, adding that all the new features were generally available.
“Multivariate time series forecasting can predict multiple time-ordered variables, where each variable depends both on its past value and the past values of other dependent variables. For example, it is used to build forecasting models to predict electricity demand in the winter considering the various sources of energy used to generate electricity,” said
Nipun Agarwal, senior vice president of research at Oracle.
In contrast to the regular practice of having a statistician trained in time-series analysis or forecasting to select the right algorithm for the desired output, AutoML’s multivariate time series forecasting automatically preprocesses the data to select the best algorithm for the ML model and automatically tunes the model, the company said.
“The HeatWave AutoML automated forecasting pipeline uses a patented technique that consists of stages including advanced time-series preprocessing, algorithm selection and hyperparameter tuning,” said Agarwal, adding that this automation can help enterprises save time and effort as they don’t need to have trained statisticians on staff.
The multivariate time series forecasting feature, according to Constellation Research Principal Analyst Holger Muller, is unique to Oracle’s MySQL HeatWave.
“Time series forecasting, multivariate or otherwise, is not currently offered as part of a single database that offers machine learning-augmented analytics. AWS, for example, offers a separate database for time series,” Muller said.
HeatWave enhances anomaly detection
Along with multivariate time series forecasting, Oracle is adding machine-learning based "unsupervised" anomaly detection to MySQL HeatWave.
In contrast to the practice of using specific algorithms to detect specific anomalies in data, AutoML can detect different types of anomalies from unlabeled data sets, the company said, adding that this feature helps enterprise users when they don’t know what anomaly types are in the dataset.
“The model generated by HeatWave AutoML provides high accuracy for all types of anomalies — local, cluster, and global. The process is completely automated, eliminating the need for data analysts to manually determine which algorithm to use, which features to select, and the optimal values of the hyperparameters,” said Agarwal.
In addition, AutoML has added a recommendation engine, which it calls recommender systems, that underpins automation for algorithm selection, feature selection, and hyperparameter optimisation inside MySQL HeatWave.
“With MySQL HeatWave, users can invoke the ML_TRAIN procedure, which automatically trains the model that is then stored in the MODEL_CATALOG. To predict a recommendation, users can invoke ML_PREDICT_ROW or ML_PREDICT_TABLE,” said Agarwal.
Business users get MySQL HeatWave AutoML console
In addition, Oracle is adding an interactive console for business users inside HeatWave.
“The new interactive console lets business analysts build, train, run, and explain ML models using the visual interface — without using SQL commands or any coding,” Agarwal said, adding that the console makes it easier for business users to explore conditional scenarios for their enterprise.
“The addition of the interactive console is in line with enterprises trying to make machine learning accountable. The console will help business users dive into the deeper end of the pool as they want to evolve into ‘citizen data scientists’ to avoid getting into too much hot water,” said Tony Baer, principal analyst at dbInsight.
The console has been made initially available for MySQL HeatWave on AWS.
Oracle also said that it would be adding support for storage on Amazon S3 for HeatWave on AWS to reduce cost as well improve the availability of the service.
“When data is loaded from MySQL (InnoDB storage engine) into HeatWave, a copy is made to the scale-out data management layer built on S3. When an operation requires reloading of data to HeatWave, such as during error recovery, data can be accessed in parallel by multiple HeatWave nodes and the data can be directly loaded into HeatWave without the need for any transformation,” said Agarwal.
MySQL Autopilot updates
The new features added to MySQL HeatWave include two new additions to MySQL Autopilot — Auto Shape prediction advisor integration with the interactive console and auto unload.
“Within the interactive console, database users can now access the MySQL Autopilot Auto shape prediction advisor that continuously monitors the OLTP workload to recommend with an explanation the right compute shape at any given time — allowing customers to always get the best price-performance,” Agarwal said.
The auto unload feature, according to the company, can recommend which tables to be unloaded based on workload history.
“Freeing up memory reduces the size of the cluster required to run a workload and saves cost,” Agarwal said, adding that both the features were in general availability.
HeatWave targets smaller data volumes
Oracle is offering a smaller shape HeatWave to attract customers with smaller sizes of data.
In contrast to the earlier size of 512GB for a standard HeatWave node, the smaller shape will have a size of 32GB with the ability to process up to 50GB for a price of $16 per month, the company said.
In addition, the company said that data processing capability for its standard 512GB HeatWave Node has been increased from 800GB to 1TB.
“With this increase and other query performance improvements, the price performance benefit of HeatWave has further increased by 15%,” said Agarwal.