GRaSP

Algorithm Introduction

Greedy relaxation of the sparsest permutation (GRaSP) algorithm [1].

Usage

from causallearn.search.PermutationBased.GRaSP import grasp

# default parameters
G = grasp(X)

# or customized parameters
G = grasp(X, score_func, depth, maxP, parameters)

# Visualization using pydot
from causallearn.utils.GraphUtils import GraphUtils
import matplotlib.image as mpimg
import matplotlib.pyplot as plt
import io

pyd = GraphUtils.to_pydot(G)
tmp_png = pyd.create_png(f="png")
fp = io.BytesIO(tmp_png)
img = mpimg.imread(fp, format='png')
plt.axis('off')
plt.imshow(img)
plt.show()

Visualization using pydot is recommended (usage example). If specific label names are needed, please refer to this usage example (e.g., GraphUtils.to_pydot(G, labels=[“A”, “B”, “C”]).

Parameters

X: numpy.ndarray, shape (n_samples, n_features). Data, where n_samples is the number of samples and n_features is the number of features.

score_func: The score function you would like to use, including (see score_functions.). Default: ‘local_score_BIC’.

maxP: Allowed maximum number of parents when searching the graph. Default: None.

parameters: Needed when using CV likelihood. Default: None.

  • parameters[‘kfold’]: k-fold cross validation.

  • parameters[‘lambda’]: regularization parameter.

  • parameters[‘dlabel’]: for variables with multi-dimensions, indicate which dimensions belong to the i-th variable.

Returns

  • G: learned general graph, where G.graph[j,i]=1 and G.graph[i,j]=-1 indicate i –> j; G.graph[i,j] = G.graph[j,i] = -1 indicates i — j.