optimask 1.3.1

OptiMask: extracting the largest (non-contiguous) submatrix without NaN

OptiMask: Efficient NaN Data Removal in Python

OptiMask is a Python package designed to facilitate the process of removing NaN (Not-a-Number) data from matrices while efficiently computing the largest (and not necessarily contiguous) submatrix without NaN values. This tool prioritizes practicality and compatibility with Numpy arrays and Pandas DataFrames.

Key Features

• Largest Submatrix without NaN: OptiMask calculates the largest submatrix without NaN, enhancing data analysis accuracy.
• Efficient Computation: With optimized computation, OptiMask provides rapid results without undue delays.
• Numpy and Pandas Compatibility: OptiMask seamlessly adapts to both Numpy and Pandas data structures.

Utilization

To employ OptiMask, install the optimask package via pip:

pip install optimask

OptiMask is also available on the conda-forge channel:

conda install -c conda-forge optimask

mamba install optimask

Usage Example

Import the OptiMask class from the optimask package and utilize its methods for efficient data masking:

from optimask import OptiMask
import numpy as np

# Create a matrix with NaN values
m = 120
n = 7
data = np.zeros(shape=(m, n))
data[24:72, 3] = np.nan
data[95, :5] = np.nan

# Solve for the largest submatrix without NaN values
rows, cols = OptiMask().solve(data)

# Calculate the ratio of non-NaN values in the result
coverage_ratio = len(rows) * len(cols) / data.size

# Check if there are any NaN values in the selected submatrix
has_nan_values = np.isnan(data[rows][:, cols]).any()

# Print or display the results
print(f"Coverage Ratio: {coverage_ratio:.2f}, Has NaN Values: {has_nan_values}")
# Output: Coverage Ratio: 0.85, Has NaN Values: False

The grey cells represent the NaN locations, the blue ones represent the valid data, and the red ones represent the rows and columns removed by the algorithm:

OptiMask’s algorithm is useful for handling unstructured NaN patterns, as shown in the following example:

Performances

OptiMask efficiently handles large matrices, delivering results within reasonable computation times:

from optimask import OptiMask
import numpy as np

def generate_random(m, n, ratio):
    """Missing at random arrays"""
    arr = np.zeros((m, n))
    nan_count = int(ratio * m * n)
    indices = np.random.choice(m * n, nan_count, replace=False)
    arr.flat[indices] = np.nan
    return arr

x = generate_random(m=100_000, n=1_000, ratio=0.02)
%time rows, cols = OptiMask(verbose=True).solve(x)
>>> 	Trial 1 : submatrix of size 37094x49 (1817606 elements) found.
>>> 	Trial 2 : submatrix of size 35667x51 (1819017 elements) found.
>>> 	Trial 3 : submatrix of size 37908x48 (1819584 elements) found.
>>> 	Trial 4 : submatrix of size 37047x49 (1815303 elements) found.
>>> 	Trial 5 : submatrix of size 37895x48 (1818960 elements) found.
>>> Result: the largest submatrix found is of size 37908x48 (1819584 elements) found.
>>> CPU times: total: 172 ms
>>> Wall time: 435 ms

Documentation

For detailed documentation, including installation instructions, API usage, and examples, visit OptiMask Documentation.

Repository Link

Find more about OptiMask on GitHub.

Citation

If you use OptiMask in your research or work, please cite it:

@software{optimask2024,
  author = {Cyril Joly},
  title = {OptiMask: NaN Removal and Largest Submatrix Computation},
  year = {2024},
  url = {https://github.com/CyrilJl/OptiMask},
}

Or:

OptiMask (2024). NaN Removal and Largest Submatrix Computation. Developed by Cyril Joly: https://github.com/CyrilJl/OptiMask