DataTransformation 0.1.1

A package for common data science operations

Data Transformation Library

This Python library provides functions commonly used for transforming data in machine learning models. The library is designed to be lightweight, efficient, and easy to use.

Functions

1. Transpose

The transpose2d function switches the axes of a 2D tensor (matrix). This operation is frequently used in data science workflows for various data manipulation tasks.

Signature:

def transpose2d(input_matrix: list[list[float]]) -> list[list[float]]:
    ...

Usage:

input_matrix = [
    [1, 2, 3],
    [4, 5, 6]
]
output_matrix = transpose2d(input_matrix)

2. Time Series Windowing

The window1d function creates a sliding window over a 1D array of data. This is particularly useful for time series analysis and modeling tasks, allowing data to be split into overlapping or non-overlapping windows for processing.

Signature:

def window1d(input_array: list | np.ndarray, size: int, shift: int = 1, stride: int = 1) -> list[list | np.ndarray]:
    ...

Usage:

input_array = [1, 2, 3, 4, 5, 6]
window_size = 3
window_shift = 1
window_stride = 1
windows = window1d(input_array, window_size, window_shift, window_stride)

3. Cross-Correlation

The convolution2d function performs cross-correlation between a 2D input matrix and a kernel matrix. Although often referred to as convolution, in deep learning, it is essentially cross-correlation. This function is commonly used in convolutional neural networks (CNNs) for feature extraction.

Signature:

def convolution2d(input_matrix: np.ndarray, kernel: np.ndarray, stride: int = 1) -> np.ndarray:
    ...

Usage:

input_matrix = np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]])
kernel = np.array([[1, 0], [0, -1]])
stride = 1
output_matrix = convolution2d(input_matrix, kernel, stride)

Project Structure

DataTransformation_library/
│
├── DataTransformation/
│   ├── __init__.py
│   ├── transpose.py
│   ├── window.py
│   └── convolution.py
│
├── tests/
│   ├── __init__.py
│   ├── test_transpose.py
│   ├── test_window.py
│   └── test_convolution.py
│
├── README.md
├── LICENSE
├── pyproject.toml
└── poetry.lock

Installation

You can install the library via pip:

pip install DataTransformation

Dependencies

• Python (>=3.6)
• NumPy (>=1.26.4)

License

This project is licensed under the MIT License. See the LICENSE file for details.