numpy_frame 0.1.0

A lightweight wrapper around numpy arrays for DataFrame-like functionality

NumpyFrame

A lightweight wrapper around numpy arrays that provides DataFrame-like functionality with minimal dependencies.

Features

• Column Selection: Access columns by name (O(1) lookup).
• Indexing & Slicing:
  • nf["col"] -> 1D array
  • nf[["col1", "col2"]] -> Sub-frame
  • nf[0:5] -> Row slice
  • nf[[0, 2], ["a", "b"]] -> Advanced rectangular indexing (Train/Test split friendly!)
• GroupBy: Split-Apply-Combine with .groupby("col").mean() or .sum().
• Fast: Built purely on top of numpy.

Installation

pip install numpy_frame

Usage

import numpy as np
from numpy_frame import NumpyFrame

# Create data
data = np.arange(12).reshape(4, 3)
cols = ["a", "b", "c"]
nf = NumpyFrame(data, cols)

# 1. Column Access
print(nf["a"])

# 2. Slicing
subset = nf[:2, ["a", "c"]] 

# 3. Advanced Indexing (e.g. random rows)
random_rows = [0, 3]
train_set = nf[random_rows, ["a", "b"]]

# 4. GroupBy
# Suppose column 'a' has groups
nf_grouped = nf.groupby("a").sum()
print(nf_grouped)

License

MIT