Description
Transform component performs feature transformation or feature engeering on the output of the Transform component. The component is used in creating Training and Inference pipelines. The Robotika Transform container is the inheritor of the Transform TFx [1] container with an extended UX.
Inputs
The Robotika container takes the python script as an input (“Transform Code location” field)
How to use
There are two ways to add the source code to Robotika container. In the field “Transform Code location” you can indicate the path to the python file which contains the preprossesing function. Please, note that the file should be located in the git repository you indicated on the first step of creating the pipelines. Another way is to load the code directly from the browser. After
The name of the function is fixed and should be “preprocessing_fn”. As an input the function takes map from feature keys to raw not-yet transformed features. As an output, the function returns a map from feature keys to transformed feature operations. The example of the code can be found below:
References
[1] TFX transform component by Google. https://www.tensorflow.org/tfx/guide/transform