Skip to content

Latest commit

 

History

History
69 lines (47 loc) · 2.55 KB

README.md

File metadata and controls

69 lines (47 loc) · 2.55 KB
Raw

Geometric Sampling Method - Python Package

The Geometric Sampling Method, introduced by Panahbehagh (2024), presents an innovative approach to finite population sampling based on a unique geometric framework. This method allows researchers to visually depict first-order inclusion probabilities (FIP) as bars on a two-dimensional graph. By adjusting the positions of these bars, users can explore a wide range of sampling designs while controlling second-order inclusion probabilities (SIP).

This package, geometric_sampling, provides tools for implementing the Geometric Sampling Method and is available on PyPI.

Features

  • Create various unequal probability sampling designs with a fixed FIP.
  • Control and explore second-order inclusion probabilities (SIP) through visual manipulation.
  • Incorporate search algorithms, such as A* and Genetic Algorithm, to optimize sampling designs for specific needs (e.g., well-spread or optimal sampling).

Installation

Install the package via pip:

pip install geometric_sampling

How to Use

The package includes core classes and methods to facilitate sampling design creation and optimization. Below is a basic example demonstrating the API.

Example Usage

import geometric_sampling as gm
import numpy as np

# Set up random generator and sample data
rng = np.random.default_rng()
N = 50  # Population size
x = rng.random(size=N)  # Auxiliary variable
n = 5  # Sample size

# Generate initial inclusion probabilities
inclusion = rng.random(N)
inclusion *= n / inclusion.sum()

# Define initial sampling design and evaluation criteria
initial_design = gm.Design(inclusion)
nht = gm.criteria.VarNHT(x, inclusion)
astar = gm.search.AStar(initial_design, nht, switch_coefficient=1)

# Display initial criteria and design
print("Initial criteria value:", astar.initial_criteria_value)
astar.initial_design.show()

# Run the A* search algorithm to optimize the design
astar.run(iterations=2000, num_new_nodes=10, max_open_set_size=10000, num_changes=1)

# Display results after optimization
print("Best criteria value:", astar.best_criteria_value)
astar.best_design.show()

Maintainers

For more details, consult the official paper: Panahbehagh, B. (2024). Geometric Sampling Method.