Great Expectations Experimental Builds

Warnings

• breaking APIs and internal implementations within `great-expectations-experimental` can change daily without notice. It is built directly from the `great_expectations` main branch.
  Fix: Avoid using this package for stable development or production environments. Always pin specific versions if testing and be prepared for frequent breaking changes.
• gotcha This package is explicitly for development and testing new features; it is not intended for production use. It may contain incomplete features, unannounced bugs, or performance issues.
  Fix: For production or stable development, use the main `great_expectations` package (`pip install great-expectations`) which follows semantic versioning and has stable releases.
• gotcha Version numbers are date-based (e.g., `0.1.20240917055`) and do not follow standard semantic versioning (MAJOR.MINOR.PATCH). This means that a 'newer' version number might not directly imply compatibility with an older one.
  Fix: When testing, always use explicit version pinning (`pip install great-expectations-experimental==0.1.20240917055`) rather than relying on `pip install --upgrade` or flexible version specifiers.
• gotcha While the package name is `great-expectations-experimental`, its Python modules are exposed under the `great_expectations` namespace (e.g., `import great_expectations as gx`). This can be confusing for new users.
  Fix: Always use `import great_expectations as gx` or `from great_expectations.<module> import <Symbol>` even when installing the experimental package.

Install

• pip install great-expectations-experimental Install latest experimental build

Imports

• great_expectations

  import great_expectations as gx

• DataContext
  wrong:

  from great_expectations_experimental.data_context import DataContext

  correct:

  from great_expectations.data_context import DataContext

  The experimental package exposes its modules under the main 'great_expectations' namespace.

Quickstart

This quickstart demonstrates how to set up an ephemeral Great Expectations DataContext using the `great-expectations-experimental` package, define a Pandas DataFrame as a data asset, and then apply and validate a couple of basic expectations. An ephemeral context is ideal for testing with this experimental package as it doesn't create persistent files.

import great_expectations as gx
import pandas as pd
import os

# Create a sample DataFrame
df = pd.DataFrame({
    "col1": [1, 2, 3, 4, 5],
    "col2": ["A", "B", "C", "D", "E"]
})

# Initialize an ephemeral DataContext
# This creates a temporary context in memory without modifying the filesystem.
context = gx.get_context(cloud_mode=False)

# Add an In-Memory Data Asset
# The name 'my_experimental_dataframe' is used to refer to this data within GX.
my_asset = context.add_pandas_dataframe_asset(
    dataframe=df, name="my_experimental_dataframe"
)

# Build a batch request for validation
batch_request = my_asset.build_batch_request()

# Get a Validator for the specified batch
validator = context.get_validator(batch_request=batch_request)

# Add an expectation: expect values in 'col1' to be between 1 and 5 (inclusive)
validator.expect_column_values_to_be_between(column="col1", min_value=1, max_value=5)

# Add another expectation: expect 'col2' to contain distinct values from a set
validator.expect_column_distinct_values_to_contain_set(column="col2", value_set=["A", "C", "E"])

# Validate the data against the defined expectations
results = validator.validate()

print("\nValidation Results:")
print(f"Overall Validation Success: {results.success}")

# You can inspect individual expectation results
for result in results.results:
    print(f"  Expectation: {result.expectation_config.expectation_type}, Success: {result.success}")