Load CSV Training Data Without Pandas in Python

This code loads a CSV file into a list of dictionaries using only the standard library, ideal for small ML training data without heavy dependencies.

Easy Python 3.6+ Aug 9, 2026 ML engineering pipelines 13 views 0 copies

csv data-loading standard-library ml-pipeline

Python code

25 lines

Python 3.6+

import csv
from pathlib import Path

def load_csv(path):
    """Load CSV file into list of dicts without pandas."""
    rows = []
    with open(path, newline='', encoding='utf-8') as f:
        reader = csv.DictReader(f)
        for row in reader:
            rows.append(dict(row))
    return rows

if __name__ == "__main__":
    data_path = Path("mock_training_data.csv")
    data_path.write_text(
        "feature1,feature2,label\n"
        "1.2,3.4,0\n"
        "5.6,7.8,1\n"
        "9.1,2.2,0\n"
    )
    data = load_csv(data_path)
    print(f"Loaded {len(data)} rows")
    for row in data:
        print(row)
    data_path.unlink()

Output

stdout

Loaded 3 rows
{'feature1': '1.2', 'feature2': '3.4', 'label': '0'}
{'feature1': '5.6', 'feature2': '7.8', 'label': '1'}
{'feature1': '9.1', 'feature2': '2.2', 'label': '0'}

How it works

The csv.DictReader reads each row as a dictionary with keys from the header row. Using dict(row) converts the OrderedDict to a plain dict. The newline='' parameter ensures proper handling of line endings across platforms. Encoding is set to UTF-8 for broad compatibility. This approach avoids pandas overhead for small datasets.

Common mistakes

Forgetting `newline=''` when opening the file, which can cause extra blank lines on Windows.
Assuming values are automatically converted to floats or ints; they remain strings.
Not closing the file if using `open` without a context manager.
Ignoring empty lines at the end of the file, which can be handled by skipping in a loop.

Variations

Use `csv.reader` and map column names manually for more control.
Use `pathlib.Path.read_text` with `csv.reader` for a single-line file read.

Real-world use cases

Loading a small static dataset for model inference in a serverless function without adding pandas to the bundle.
Reading training features from a CSV exported by a labeling tool in a quick prototype script.
Parsing configuration-like CSV files in a data pipeline where pandas is not installed.

Load CSV Training Data Without Pandas in Python

Python code

Output

How it works

Common mistakes

Variations

Real-world use cases

More from ML engineering pipelines

Tutorials

Quizzes

Python code

Output

How it works

Common mistakes

Variations

Real-world use cases

More from ML engineering pipelines

Keep learning

Tutorials

Quizzes