lealre / one-billion-rows-challenge

Reading One Billion Rows in Python

One Billion Rows Challenge

This repository represents Project 01 from the Data Engineering BOOTCAMP, where the objective is to compare different ways of reading one billion rows in Python. Upon cloning the original repository, the objective is to implement script modifications based on the concepts learned in previous bootcamp classes, such as type hints and functions.

From the original repository, a decorator function was implemented to measure the execution time of each script function responsible for reading the file. Minor structural adjustments, such as introducing a type hint in the path variable, were made to the scripts. Additionally, the Dask script was intentionally excluded, narrowing the comparison to Python (native structure), Pandas, Polars, and DuckDB.

Results

Machine specifications:

• 11th Gen Intel(R) Core(TM) i5-1135G7 @ 2.40GHz 2.42 GHz
• 16G RAM

Enviroment	Execution Time
Python	46 min 37 s
Python + Pandas	6 min 36 s
Python + Polars	38 s
Python + Duckdb	33 s

How To Run

How to execute this repository in a bash terminal using pyenv and poetry.

Before you begin, ensure that measurements.txt is inside .gitignore

1. Clone the repository:

git clone https://github.com/lealre/one-billion-rows-challenge.git

2. Enter the folder:

cd one-billion-rows-challenge

3. Set Python version using pyenv:

pyenv local 3.12.1

4. Set poetry to use Python 3.12.1:

poetry env use 3.12.1

5. Install dependencies and activate the virtual environment:

poetry install --no-root

poetry lock --no-update

poetry shell

6. Create the measurements.txt file (takes some time to create):

python src/create_measurements.py

7. Read the file using specific script:

python src/using_python.py 

python src/using_pandas.py 

python src/using_polars.py 

python src/using_duckdb.py 

---

THIS IS A CLONE FROM THIS REPO! The scripts designed to read the file are not authored by me.