Polars vs pandas

gistfile1.py

import pandas as pd
import polars as pl
import time

# Generate a sample CSV file for the test
data = {
    "col1": range(1, 1000001),
    "col2": range(1000000, 0, -1),
    "col3": ["text"] * 1000000
}
df = pd.DataFrame(data)
df.to_csv("sample.csv", index=False)

# Read and manipulate data using pandas
start_time = time.time()
pandas_df = pd.read_csv("sample.csv")
pandas_df['col4'] = pandas_df['col1'] + pandas_df['col2']
pandas_sum = pandas_df['col4'].sum()
pandas_time = time.time() - start_time

print(f"Pandas operation took {pandas_time:.4f} seconds")

# Read and manipulate data using Polars
start_time = time.time()
polars_df = pl.read_csv("sample.csv")
polars_df = polars_df.with_columns((polars_df['col1'] + polars_df['col2']).alias('col4'))
polars_sum = polars_df['col4'].sum()
polars_time = time.time() - start_time

print(f"Polars operation took {polars_time:.4f} seconds")

# Verify that the results are the same
assert pandas_sum == polars_sum, "The results do not match!"

# Output comparison
print(f"Pandas took {pandas_time:.4f} seconds, Polars took {polars_time:.4f} seconds")

Pandas operation took 0.1143 seconds
Polars operation took 0.0451 seconds
Pandas took 0.1143 seconds, Polars took 0.0451 seconds