Pawel+Jankiewicz.png

Pawel Jankiewicz


Pawel Jankiewicz is a Lead Data Scientist at Growbots, an AI-powered outbound sales platform. He has a background in Finance and Banking, working for 6 years in one of the biggest banks in Poland as a reporting specialist. After repeated success in Data Science competitions, he switched careers. He works on Natural Language Processing analytics. His favorite tools include ensembling and tree-based methods.
Pawel holds a Master's degree in Investment Banking from Warsaw School of Economics. He enjoys Kaggle, and has achieved 8 honors at competitions.
https://www.linkedin.com/in/pjankiewicz/
https://github.com/pjankiewicz

Advanced machine learning: Model Pipelines

Learning outcomes:

  • How to weight, transform, combine, or drop features
  • How to represent transformations, models, parameters, and the results of a run, so they can be easily managed
  • What feature transformations add the most performance, and how they interact with the rest of the pipeline

Once you start fitting models, methodology matters. It is easy simply to pile up complexity, without managing it. Fortunately, we now have best practices (and libraries) that make it easy to iterate over preprocessing, model families, and parameters.

Content

Using best practices for model pipelines, Implementing features to interact a robust pipeline


Pre-requisites

  • Basic to intermediate knowledge of machine learning (that is, you have run models and optimized parameters, but are sure there must be a better way to do it).