Pyarrow

• Pyarrow for faster data reads
  • pip install pyarrow
• video format of this article https://youtu.be/gFd4I1oXG8E

# Create a dummy dataset:
import random
import string
import numpy as np
import pandas as pd
import pyarrow as pa
import pyarrow.csv as csv
from datetime import datetime

def gen_random_string(length: int = 32) -> str:
    return ''.join(random.choices(
        string.ascii_uppercase + string.digits, k=length)
    )

dt = pd.date_range(
    start=datetime(2000, 1, 1),
    end=datetime(2021, 1, 1),
    freq='min'
)

np.random.seed = 42
df_size = len(dt)
print(f'Dataset length: {df_size}')


df = pd.DataFrame({
    'date': dt,
    'a': np.random.rand(df_size),
    'b': np.random.rand(df_size),
    'c': np.random.rand(df_size),
    'd': np.random.rand(df_size),
    'e': np.random.rand(df_size),
    'str1': [gen_random_string() for x in range(df_size)],
    'str2': [gen_random_string() for x in range(df_size)]
})
# 1 test baseline speed of pandas
## even being kind by using the compression parameter

df.to_csv('csv_pandas.csv.gz', index=False, compression='gzip')

# Check read times

df1 = pd.read_csv('csv_pandas.csv')

# **NOTE:** Here’s a thing you should know about PyArrow — it can’t handle datetime columns. You’ll have to convert the date attribute to a timestamp. Here’s how:

df_pa = df.copy()
df_pa['date'] = df_pa['date'].values.astype(np.int64) // 10 ** 9
df_pa_table = pa.Table.from_pandas(df_pa)
csv.write_csv(df_pa_table, 'csv_pyarrow.csv')

# Adding compression

with pa.CompressedOutputStream('csv_pyarrow.csv.gz', 'gzip') as out:
    csv.write_csv(df_pa_table, out)

# You can read both compressed and uncompressed dataset with the csv.read_csv() function:

df_pa_1 = csv.read_csv('csv_pyarrow.csv')
df_pa_2 = csv.read_csv('csv_pyarrow.csv.gz')

# Both will be read in the pyarrow.Table format, so use the following command to convert them to a Pandas DataFrame:

df_pa_1 = df_pa_1.to_pandas()

---

• Reference:
  • Write Data This Alternative Is 7 Times Faster