Preprocessing data with MLOps Codex

Now you can preprocess multiples data with MLOps Codex using the the new preprocessing.

With the mlops_codex.preprocessing.MLOpsPreprocessingClient class give you the interaction between your code and our environment.

Hosting a preprocess script

Let’s get started. First import and connect to MLOps:

from mlops_codex.preprocessing import MLOpsPreprocessingClient

client = MLOpsPreprocessingClient()

Prepare your data. In this case, you must do the following:

Note the data structure must be as above

Next, send the data to our server, by using the .create() method. Check the parameters doc.

To do it, you must do as following:

The wait_read and host are optional parameters. They control the initial state of the preprocessing object.

Creating an execution

To create a new execution, do as the following code:

Download the result of your preprocessing

The result of a preprocessing execution is a preprocessed_data.parquet. To download the result, check the example below:

run.download()