Skip to content
A data transformation package for deep learning with Autonomio, Keras and TensorFlow.
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.


Data preparation for deep learning

Talos Travis Talos Coveralls

WrangleKey FeaturesInstallSupportIssuesLicenseDownload

Wrangle provides the building blocks for entirely avoiding redundant, easy-to-automate, data preparation tasks.



Wrangle dramatically simplifies 95% of data preparation tasks involved in advanced deep learning practice and provides the required building blocks for near-future automated machine intelligence workflows. Wrangle is created to solve the problem of avoiding beneficial workflow steps due to complexity, cognitive overhead, and the anxiety that comes with it.

Key Features

Because of the large number of functions, many of which are frequently used in common deep learning data preparation workflows, Wrangle is notably focused on namespace. All functions are named in a way where the name explains exactly what can be expected in terms of capability. Let's dissect a few as an example:

In col_to_binary col refers to what is being processed, in this case a column of a dataframe. to refers to the particular process, in this case a conversion. binary refers to the output. In this case a given column in a dataframe is converted into binary values. For example, a continuous column is converted to binary classes based on if the values are below or above mean value. Similarly array_reshape_conv1d can be understood as taking in an array, and reshaping it to conv1d layer required shape.

Wrangle key features include:

  • Resampling
  • Transformation
  • Renaming
  • Grouping
  • Merging
  • Correlations
  • Reshaping
  • Cleaning

Wrangle works on Linux, Mac OSX, and Windows systems.


Stable version:

pip install wrangle

Daily development version:

pip install git+


If you want ask a "how can I use Wrangle to..." question, the right place is StackOverflow.

If you found a bug or want to suggest a feature, check the issues or create a new issue.


MIT License

You can’t perform that action at this time.