Plan Your Deployment

This solution uses a best-practice deployment pattern for running an AI-powered sales and revenue forecasting solution on Oracle Cloud Infrastructure (OCI).

Although many OCI services can be replaced by open-source components, using OCI services offers significant advantages, such as a simplified code base and smooth, seamless deployments through deep integrations between the listed services. For example, you can make a connection between Oracle Cloud Infrastructure Data Science and Oracle Autonomous Data Warehouse with a single line of code. Similarly, Oracle Cloud Infrastructure Data Integration provides built-in connectors for a large number source and targets so that no custom code is needed.

Description of the illustration ai-forecasting-functional.png

ai-forecasting-functional-oracle.zip

Oracle Cloud Infrastructure Data Science is a fully managed service that empowers data scientists to seamlessly build, train, and deploy machine learning models. It offers a collaborative environment with Jupyter-based notebooks, support for popular machine learning frameworks, and scalable compute resources, including GPU support for high-performance workloads. With built-in tools for model tracking, versioning, and deployment, Oracle Cloud Infrastructure Data Science simplifies the entire machine learning lifecycle.

In addition, Oracle Cloud Infrastructure Data Science, provides AI Forecaster libraries and the AI Forecast Operator to generate forecasts for future trends by using historical time series data.

The AI Forecast Operator simplifies and accelerates the process by automating model selection, hyperparameter tuning, and feature identification for a specific prediction task. Users can see and interpret results by using helpful visualizations generated by the forecasting process:.

Description of the illustration forecast-operator-display.png

The following are some implementation alternatives:

• Instead of Oracle Digital Assistant, you can also create a custom chatbot by using Oracle APEX Application Development or by using an open-source application development tool, such as StreamLit.
• You can use out-of-the-box large language models (LLMs) offered in Oracle Cloud Infrastructure Generative AI, such as Cohere, and LLaMA or use other publicly available LLMs.
• Although Oracle recommends using the Oracle Autonomous Database, you can use any relational database.

Use the following high-level steps to deploy the solution:

1. Provision Oracle Cloud Infrastructure Data Integration by using the OCI console and the intuitive user interface to create workspaces, and underlying projects with data sources and data ingestion pipelines.
2. Deploy Oracle Autonomous Database for analytics and data warehousing with a few clicks in the OCI console. You can choose shared or dedicated deployment options based on your preference for isolation of workloads. Deployment options include support for Exadata Cloud@Customer, OCI Dedicated Region, and Multi-cloud. Select Autoscaling to maintain continuity during traffic spikes and other fluctuations that impact workload volumes.
3. Provision Oracle Cloud Infrastructure Data Science by using the OCI console. OCI provides out-of-the box starter code for building Forecasting Operators and the associated Model Endpoint for interpreting results.
4. Provision Oracle Digital Assistant to provide chat and voice interfaces and to provide advanced natural language capabilities that surpass those of simple chatbots. This product can support complex business workflows, however for this solution, a small subset of functionality is used to invoke REST services and to present a friendly use interface.
5. Large language models from Oracle Cloud Infrastructure Generative AI enrich the user-provided question and the associated response with additional context.