Install GX

GX 1.0 is a Python library. Follow the instructions in this guide to install GX in your local Python environment, or as a notebook-scoped library in hosted environments such as Databricks or EMR Spark clusters.

Prerequisites

• Python version 3.8 to 3.11
• Recommended. A Python virtual environment
• Internet access
• Permissions to download and install packages in your environment

Install the GX Python library

Local

GX 1.0 is a Python library and as such can be used with a local Python installation to access the functionality of GX through Python scripts.

Installation and setup

1. Optional. Activate your virtual environment.

   If you created a virtual environment for your GX Python installation, browse to the folder that contains your virtual environment and run the following command to activate it:

   Terminal input

   source my_venv/bin/activate

2. Ensure you have the latest version of pip:

   Terminal input

   python -m ensurepip --upgrade

3. Install the GX 1.0 library:

   Terminal input

   pip install great_expectations

4. Verify that GX installed successfully with the terminal command:

   Terminal input

   great_expectations --version

   If GX was successfully installed, the following output appears:

   Terminal output

   great_expectations, version 1.0.0a4

Hosted environment

Hosted environments such as EMR Spark or Databricks clusters do not provide a filesystem to install your GX instance. Instead, you must install GX in memory using the Python-style notebooks available on those platforms.

EMR Spark notebook

Use the information provided here to install GX on an EMR Spark cluster and instantiate a Data Context without a full configuration directory.

Additional prerequisites

• An EMR Spark cluster.
• Access to the EMR Spark notebook.

Installation and setup

1. To install GX on your EMR Spark cluster copy this code snippet into a cell in your EMR Spark notebook and then run it:

   Python

   sc.install_pypi_package("great_expectations")

2. Create an in-code Data Context. See Instantiate an Ephemeral Data Context.

3. Copy the Python code at the end of How to instantiate an Ephemeral Data Context into a cell in your EMR Spark notebook, or use the other examples to customize your configuration. The code instantiates and configures a Data Context for an EMR Spark cluster.

4. Execute the cell with your Data Context initialization and configuration.

5. Run the following command to verify that GX was installed and your in-memory Data Context was instantiated successfully:

   Python

      context.list_datasources()

Databricks notebook

Databricks is a web-based platform that automates Spark cluster management.

To avoid configuring external resources, you'll use the Databricks File System (DBFS) for your Metadata Stores and Data Docs store.

DBFS is a distributed file system mounted in a Databricks workspace and available on Databricks clusters. Files on DBFS can be written and read as if they were on a local filesystem by adding the /dbfs/ prefix to the path. It also persists in object storage, so you won’t lose data after terminating a cluster. See the Databricks documentation for best practices, including mounting object stores.

Additional prerequisites

• A complete Databricks setup, including a running Databricks cluster with an attached notebook
• Access to DBFS

Installation and setup

1. Run the following command in your notebook to install GX as a notebook-scoped library:

   Terminal input

   %pip install great-expectations

   A notebook-scoped library is a custom Python environment that is specific to a notebook. You can also install a library at the cluster or workspace level. See Databricks Libraries.

2. Run the following command to import the Python configurations you'll use in the following steps:

   Python

   import great_expectations as gx
   from great_expectations.checkpoint import Checkpoint
   from great_expectations.core.expectation_suite import ExpectationSuite

3. Run the following code to specify a /dbfs/ path for your Data Context:

   Python

   context_root_dir = "/dbfs/great_expectations/"

4. Run the following code to instantiate your Data Context:

   Python

   context = gx.get_context(context_root_dir=context_root_dir)

GX Cloud

GX Cloud provides a web interface for using GX to validate your data without creating and running complex Python code. However, GX 1.0 can connect to a GX Cloud account if you want to customize or automate your workflows through Python scripts.

Installation and setup

To deploy a GX Agent, which serves as an intermediary between GX Cloud's interface and your organization's data stores, see Connect GX Cloud. The GX Agent serves all GX Cloud users within your organization. If a GX Agent has already been deployed for your organization, you can use the GX Cloud online application without further installation or setup.

To connect to GX Cloud from a Python script utilizing a local installation of GX instead of the GX Agent, see Connect to an existing Data Context.

Next steps

• Install additional dependencies
• Manage Data Contexts