csr-stat

Package for introductory analysis of Spatial Point Processes, including basic utility functions, summary statistics and metrics for 2D point processes. To be primarily used to test the hypothesis of CSR.

---

Installation

git clone https://github.com/chancehaycock/csrstat.git

# Create conda environment if you don't already have one
conda create -n my_env python==3.7
conda activate my_env
conda install pytest
cd csrstat
pip install .

Check installation with

pytest test/

---

Included Summary Statistics

These functions calculate the associated metric of an observed Point Process PP. When restrict_domain is a valid argument, the metric is calculated on a subset of the original domain, in a first attempt to account for edge effects.

F_hat(PP, r, num_sampled_points=1000, plot=False)

G_hat(PP, r, plot=False)

J_hat(PP, r, num_F_sampled_points=1000, plot=False)

K_hat(PP, r, restrict_domain=False, plot=False)

L_hat(PP, r, restrict_domain=False, plot=False)

O_hat(PP, r, bandwidth=0.1, restrict_domain=False, kernel="BK", plot=False)

PC_hat(PP, r, bandwidth=0.1, restrict_domain=False, kernel="BK", plot=False)

Additional Fast Functions

Good for simulation inference.

K_hat_fast(PP, r, plot=False)

L_hat_fast(PP, r, plot=False)

Simulation

Any of the above summary statistic functions can be simulated by calling the following. Resulting max/min envelopes will then be plotted and can be compared against a given realisation. Currently, only the default arguments of a summary statistic function can be used during simulation.

simulate_summary_statistic(PP, r, summary_func, n_sims=100, plot=False)

Real-Valued Hypothesis Test

CSR_hypothesis_test_dmin(observed_PP, significance_level, n_sim=500, plot_dist=False)

Testing

pytest test/utils_test.py