1000x faster data manipulation: vectorizing with Pandas and Numpy

  Рет қаралды 56,014

PyGotham 2019

PyGotham 2019

Күн бұрын

Speaker: Nathan Cheever
The data transformation code you're writing is correct, but potentially
1000x slower than it needs to be! In this talk, we will go over multiple
ways to enhance a data transformation workflow with Pandas and Numpy by
showing how to replace slower, perhaps more familiar, ways of operating on
Pandas data frames with faster-vectorized solutions to common use cases
like:
* if-else logic in applied row-wise functions
* dictionary lookups with conditional logic
* Date comparisons and calculations
* Regex and string column manipulation
* and others! ...
without needing a beefier computer, writing Cython, or other libraries
outside the Pandas ecosystem.

Пікірлер
Losing your Loops Fast Numerical Computing with NumPy
30:31
PyCon 2015
Рет қаралды 80 М.
World’s Deadliest Obstacle Course!
28:25
MrBeast
Рет қаралды 92 МЛН
Sprinting with More and More Money
00:29
MrBeast
Рет қаралды 184 МЛН
Они убрались очень быстро!
00:40
Аришнев
Рет қаралды 3,3 МЛН
Make Python code 1000x Faster with Numba
20:33
Jack of Some
Рет қаралды 439 М.
Speed Up Your Pandas Dataframes
11:15
Rob Mulla
Рет қаралды 68 М.
Основы NumPy Python | Массивы, Матрицы И Операции Над Ними
38:26
PyLounge - программирование на Python и всё о IT
Рет қаралды 101 М.
Static Typing in Python
28:19
PyGotham 2019
Рет қаралды 4,5 М.
Three ways to optimize your Pandas data frame's memory footprint
13:37
Python and Pandas with Reuven Lerner
Рет қаралды 2,1 М.
Python dataclasses will save you HOURS, also featuring attrs
8:50
This INCREDIBLE trick will speed up your data processes.
12:54
Rob Mulla
Рет қаралды 257 М.
Miguel Raz Guzmán Macedo - Portable SIMD tricks for fun and profit
9:53
Scientific Computing in Rust
Рет қаралды 2 М.
World’s Deadliest Obstacle Course!
28:25
MrBeast
Рет қаралды 92 МЛН