End-to-end samples: Industry accelerators

The industry accelerators that are provided by IBM are a set of end-to-end solutions that you can run as examples or customize them to address common business issues.

Overview of industry accelerators

An industry accelerator is designed to help you solve a specific business problem, whether it's predicting how customers will respond to a promotion or understanding customer attrition. Browse the Accelerators catalog for the Cloud Pak for Data industry accelerators and download the ones that you want.

Most accelerators include a Sample project with everything that you need to analyze data, build a model, and display results. The sample projects include detailed instructions, datasets, Jupyter notebooks, models, and R Shiny applications. Most accelerators also include business terms for data governance. Use these sample projects as templates for your own data science needs to learn specific techniques, or to demonstrate the capabilities of Watson Studio and other IBM watsonx services.

Most accelerators also include a Business glossary that consists of terms and categories for data governance. The terms and categories provide meaning to the accelerator and act as the information architecture for the accelerator.

The process of using an industry accelerator is illustrated in this graphic.

Industry accelerator process

Sample project for an accelerator

When you download an accelerator, you get a project that contains the assets that you need to build and train the models that are associated with the accelerator.

Audience

Data scientists or business analysts who analyze data and build models to solve business problems.

Contents

A typical accelerator contains the following items:

  • A readme file that provides instructions.
  • CSV files that contain the sample data.
  • Python notebooks to train and score the models and associated Python scripts to prepare and transform the data for modeling. The notebooks create API endpoints to expose the output for the R Shiny application.
  • Machine learning models that are designed to help you find answers to the business problems described by the accelerator.
  • (Some accelerators) An interactive dashboard to show the results of the model.

Requirements

All accelerators that have sample projects require the Watson Studio service and one or more of these services:

  • Watson Machine Learning
  • IBM Knowledge Catalog
  • RStudio Runtime
  • Decision Optimization
  • SPSS Modeler
  • Watson OpenScale

To check which services you have, see Finding services.

Process overview

Each accelerator provides detailed instructions. These steps provide an overview of the process for sample projects:

  1. Import the project. See Importing a project.
  2. Follow the instructions in the readme file to run the notebooks and complete other tasks.

Next steps

You can use the sample project as a template by adding your own data and following the same steps to go from data to deployed model. You might need to explore and cleanse your data. Because your data and schema are likely different from the sample data, the patterns that you find in your data might not match the patterns in the sample data. You can use the examples and code to adapt the model for your data and to retrain the model with your data.

To use a sample project as a template for your own data, follow these general steps:

  1. Add your data to the project.
  2. If necessary, cleanse or shape your data.
  3. Update and retrain the model with your data.

Business glossary

A business glossary helps you describe your data with a standard vocabulary. You import categories and business terms as governance artifacts.

Audience

Data stewards and other users who are responsible for creating governance artifacts to govern data.

Contents

A set of business terms to describe data, with logical relationships between terms. See Business terms.

A category with the same name as the accelerator in which to organize the terms. See Categories.

Requirements

All accelerators that have a business glossary require the IBM Knowledge Catalog service.

You must have the Manage governance categories permission. To see which permissions you have, see Checking your permissions.

Process overview

Each accelerator provides detailed instructions that you can access after you download the accelerator. These general steps provide an overview of the process:

  1. Import the category.
  2. Import the business terms.
  3. Publish the business terms so that they are available in all catalogs and during automated discovery.
  4. Assign business terms to columns in datasets with one of these methods:
    • Add business terms to a data asset in a catalog on the asset's Overview page. See Editing asset properties.
    • Some accelerators provide a Jupyter Notebook that assigns terms to the datasets included in the sample project.