Documentation | Release Notes | Examples | Tutorial
The dbldatagen
Databricks Labs project is a Python library for generating synthetic data within the Databricks
environment using Spark. The generated data may be used for testing, benchmarking, demos, and many
other uses.
It operates by defining a data generation specification in code that controls how the synthetic data is generated. The specification may incorporate the use of existing schemas or create data in an ad-hoc fashion.
It has no dependencies on any libraries that are not already installed in the Databricks runtime, and you can use it from Scala, R or other languages by defining a view over the generated data.
It supports:
- Generating synthetic data at scale up to billions of rows within minutes using appropriately sized clusters
- Generating repeatable, predictable data supporting the need for producing multiple tables, Change Data Capture, merge and join scenarios with consistency between primary and foreign keys
- Generating synthetic data for all of the Spark SQL supported primitive types as a Spark data frame which may be persisted, saved to external storage or used in other computations
- Generating ranges of dates, timestamps, and numeric values
- Generation of discrete values - both numeric and text
- Generation of values at random and based on the values of other fields
(either based on the
hash
of the underlying values or the values themselves) - Ability to specify a distribution for random data generation
- Generating arrays of values for ML-style feature arrays
- Applying weights to the occurrence of values
- Generating values to conform to a schema or independent of an existing schema
- use of SQL expressions in synthetic data generation
- plugin mechanism to allow use of 3rd party libraries such as Faker
- Use within a Databricks Delta Live Tables pipeline as a synthetic data generation source
- Generate synthetic data generation code from existing schema or data (experimental)
- Use of standard datasets for quick generation of synthetic data
Details of these features can be found in the online documentation - online documentation.
Please refer to the online documentation for details of use and many examples.
Release notes and details of the latest changes for this specific release can be found in the GitHub repository here
Use pip install dbldatagen
to install the PyPi package.
Within a Databricks notebook, invoke the following in a notebook cell
%pip install dbldatagen
The Pip install command can be invoked within a Databricks notebook, a Delta Live Tables pipeline and even works on the Databricks community edition.
The documentation installation notes contains details of installation using alternative mechanisms.
The Databricks Labs Data Generator framework can be used with Pyspark 3.1.2 and Python 3.8 or later. These are compatible with the Databricks runtime 10.4 LTS and later releases. For full Unity Catalog support, we recommend using Databricks runtime 13.2 or later (Databricks 13.3 LTS or above preferred)
For full library compatibility for a specific Databricks Spark release, see the Databricks release notes for library compatibility
When using the Databricks Labs Data Generator on "Unity Catalog" enabled Databricks environments,
the Data Generator requires the use of Single User
or No Isolation Shared
access modes when using Databricks
runtimes prior to release 13.2. This is because some needed features are not available in Shared
mode (for example, use of 3rd party libraries, use of Python UDFs) in these releases.
Depending on settings, the Custom
access mode may be supported.
The use of Unity Catalog Shared
access mode is supported in Databricks runtimes from Databricks runtime release 13.2
onwards.
See the following documentation for more information:
To use the data generator, install the library using the %pip install
method or install the Python wheel directly
in your environment.
Once the library has been installed, you can use it to generate a data frame composed of synthetic data.
The easiest way to use the data generator is to use one of the standard datasets which can be further customized for your use case.
import dbldatagen as dg
df = dg.Datasets(spark, "basic/user").get(rows=1000_000).build()
num_rows=df.count()
You can also define fully custom data sets using the DataGenerator
class.
For example
import dbldatagen as dg
from pyspark.sql.types import IntegerType, FloatType, StringType
column_count = 10
data_rows = 1000 * 1000
df_spec = (dg.DataGenerator(spark, name="test_data_set1", rows=data_rows,
partitions=4)
.withIdOutput()
.withColumn("r", FloatType(),
expr="floor(rand() * 350) * (86400 + 3600)",
numColumns=column_count)
.withColumn("code1", IntegerType(), minValue=100, maxValue=200)
.withColumn("code2", IntegerType(), minValue=0, maxValue=10)
.withColumn("code3", StringType(), values=['a', 'b', 'c'])
.withColumn("code4", StringType(), values=['a', 'b', 'c'],
random=True)
.withColumn("code5", StringType(), values=['a', 'b', 'c'],
random=True, weights=[9, 1, 1])
)
df = df_spec.build()
num_rows=df.count()
Refer to the online documentation for further examples.
The GitHub repository also contains further examples in the examples directory.
The dbldatagen
package is intended to be compatible with recent LTS versions of the Databricks runtime, including
older LTS versions at least from 10.4 LTS and later. It also aims to be compatible with Delta Live Table runtimes,
including current
and preview
.
While we don't specifically drop support for older runtimes, changes in Pyspark APIs or
APIs from dependent packages such as numpy
, pandas
, pyarrow
, and pyparsing
make cause issues with older
runtimes.
By design, installing dbldatagen
does not install releases of dependent packages in order
to preserve the curated set of packages pre-installed in any Databricks runtime environment.
When building on local environments, the build process uses the Pipfile
and requirements files to determine
the package versions for releases and unit tests.
Please note that all projects released under Databricks Labs
are provided for your exploration only, and are not formally supported by Databricks with Service Level Agreements
(SLAs). They are provided AS-IS, and we do not make any guarantees of any kind. Please do not submit a support ticket
relating to any issues arising from the use of these projects.
Any issues discovered through the use of this project should be filed as issues on the GitHub Repo.
They will be reviewed as time permits, but there are no formal SLAs for support.
Issues with the application? Found a bug? Have a great idea for an addition? Feel free to file an issue.