Oracle Fusion Analytics provides ready-to-use machine learning (ML) for specific business processes. Additionally, users can leverage the same Fusion Analytics platform services to build their own ML-powered use cases or easily apply self-service ML and predictive analytics with a few clicks, supporting business users, analysts, and data scientists.
Oracle Fusion Analytics continues to expand its library of ready-to-use machine learning for specific business processes so that users can uncover their own insights and results. Here are some samples:
Predicts risk of a customer paying late or an invoice being paid late. Helps prioritize collections and improve cash flow.
Enables detection and monitoring of adverse impact indicators in hires, terminations, and promotions by gender and ethnicity.
Oracle Fusion Analytics also includes an enterprise-scale machine learning platform for data scientists to run machine learning within the database, where the data resides. This platform, called Machine Learning in Oracle Database, includes more than thirty ML algorithms to provide no-code automated machine learning. Machine Learning in Oracle Database provides natural interfaces for popular programming languages used in data science, such as SQL, R, and Python.
Most importantly, Fusion Analytics provides citizen data scientists and analysts a self-service method for accessing these models from the central repository and easily executing them on their own datasets to generate predictions.
Leveraging the capabilities of the underlying Oracle Analytics Cloud, anyone can apply machine learning and predictive analytics to quickly detect anomalies and predict outcomes.
The Explain capability enables you to examine any dataset to quickly identify meaningful business drivers and data anomalies with only a few clicks. Get automatic visualizations in return to jump-start new, deeper analyses.
Easily apply prebuilt advanced analytics with a few clicks.
Oracle Fusion Analytics includes various machine learning algorithms to help build and train predictive models to predict a target value or identify classes of records—no coding required. Examples of available algorithm types include classification and regression trees (CART), logistic regression, and k-means. After the predictive model has been trained, anyone can apply it to any dataset.