How to Use to Track Experimentation: An Example With Structured Data and XGBoost

1 min
Kamil Kaczmarek
21st April, 2022

What will you learn?

How you can use Neptune to track experimentation on the example with structured data and XGBoost.

  • 00:30

    How to add Neptune to your training scripts?

  • 00:45

    The model training script

  • 01:03

    The neptune.init method

  • 01:33

    The run object is a handler to the run in Neptune

  • 01:48

    Neptune + XGBoost integration

  • 02:33

    Pass the Neptune callback to the XGBoost train to log metadata automatically during training

  • 03:04

    Download the trained model from Neptune

  • 03:53

    How does logged metadata look like in Neptune?

  • 04:36

    The dashboard can be composed of multiple metrics logged to the run

  • 05:27

    Make use of the hierarchical structure of the run

  • 05:46

    Log images, such as feature importances of visualized trees

  • 06:14

    The metrics tracked during the run are visualized as interactive charts

  • 06:31

    Use the dashboard to analyze the run from multiple points of view

  • 06:51

    The dashboards are interactive and can be composed of multiple widgets

Important: This video was created in November 2021. For the most up-to-date code examples, please refer to the Neptune docs. 

Other useful resources

Learn also about API reference: XGBoost integration.

Couldn’t find the use case you were looking for?

Just get in touch, and our ML team will create a custom demo for you.