Module 2: NumPy “where” and NumPy “select” to Boost Pandas “apply” statements and ndarray performance

Bob Chesebrough and Praveen Kundurthy, Intel
ALP Graphic Module 2

This is the second module in the newest Aurora Learning Paths series - Accelerate Python Loops with the Intel AI Analytics Toolkit. Ignite performance of common Python, ndarray & pandas “apply” constructs by using NumPy, SciPy, and Pandas powered by oneAPI. This webinar and workshop series will go into detail about how to apply key Intel architectural innovations and libraries via smart application of NumPy techniques to achieve amazing performance gains.  We’ll delve into NumPy powered by oneAPI where and select clauses to allow us to vectorize loops with conditional logic, as well as applications to Pandas Apply statements and exploration of higher level abstractions using SciPy. Learn how to achieve performance gains by replacing Python loop centric or list comprehension applications with smarter equivalents that are more maintainable, more efficient, and much faster on current and future innovations in Intel hardware and oneAPI software libraries!

 Learning Outcomes:

  • Apply NumPy constructs as a Python loop replacement strategy that  improves readability, maintainability, performs fasts on current hardware and readies code for future HW & SW accelerations that Intel builds into silicon and which are exposed via NumPy
  •  Apply NumPy sorting, aggregations, reductions, broadcasting, and “where” and “select” clauses to significantly accelerate your Python code
  • Apply NumPy where and select clauses to improve performance of Pandas apply statements impacted by conditional logic
  • Understand the value of NumPy to accelerate Python loops