Machine Learning

The diagram illustrates the integration of Machine Learning (ML) Data Science with a BDC IoT Cloud ecosystem, involving various components responsible for data processing, model training, and IoT operations.

Machine Learning Model Process

The machine learning model process involves data engineering, where raw data is prepared and then stored in a repository. Data scientists build and train machine learning models using this data. The trained ML models generate inferences based on IoT data, enabling the BDC IoT Cloud to manage and automate IoT devices through the Rules-Engine. Salesforce provides external business integration to align IoT operations with business processes.

Data Pipeline (Data Engineering)

• Preprocess Data: Raw data is collected and preprocessed for quality and format consistency by a Data Engineer.

• Data Repository: Once preprocessed, the data is stored in a repository, which serves as input for the Model Pipeline.

Model Pipeline (Data Science)

• Feature Engineering and Training: The Data Scientist processes data from the repository, applying feature engineering techniques and training machine learning models.
• Evaluation: After training, the model undergoes evaluation to ensure its performance on test datasets.

Release Pipeline

• The Release Pipeline handles the deployment of the trained and evaluated model. This process includes:

  • Deploy: Deploying the model to production infrastructure.
  • Approve: Stakeholder approval.
  • Profile: Analyzing the model’s performance.
  • Validate: Ensuring the model meets functional and performance requirements.
  • Package: Packaging the final model for release.

• The model is stored in the Model Registry, which serves as a centralized repository for managing and tracking machine learning models. The Model Registry ensures that every version of the model, along with its metadata (such as deployment versions, and associated configurations) is easily accessible for future use or updates.

BDC Data Cloud and Salesforce

• The ML Model developed in the Machine Learning Data Science section produces Model Inferences based on IoT data. These inferences are fed into the BDC IoT Rules-Engine, where they help automate processes within the BDC IoT Cloud by applying decision-making logic.

• The interation between Salesforce and BDC IoT Cloud helps to synchronize customer data, device management, and analytics with business processes.