hidimstat.model_x_knockoff¶
- hidimstat.model_x_knockoff(X, y, fdr=0.1, offset=1, method='equi', statistics='lasso_cv', shrink=False, centered=True, cov_estimator='ledoit_wolf', verbose=False, memory=None, n_jobs=1, seed=None)¶
Model-X Knockoff inference procedure to control False Discoveries Rate, based on Candes et al.[1]
- Parameters:
- X2D ndarray (n_samples, n_features)
design matrix
- y1D ndarray (n_samples, )
response vector
- fdrfloat, optional
desired controlled FDR level
- offsetint, 0 or 1, optional
offset to calculate knockoff threshold, offset = 1 is equivalent to knockoff+
- methodstr, optional
knockoff construction methods, either equi for equi-correlated knockoff or sdp for optimization scheme
- statisticsstr, optional
method to calculate knockoff test score
- shrinkbool, optional
whether to shrink the empirical covariance matrix
- centeredbool, optional
whether to standardize the data before doing the inference procedure
- cov_estimatorstr, optional
method of empirical covariance matrix estimation
- seedint or None, optional
random seed used to generate Gaussian knockoff variable
- Returns:
- selected1D array, int
vector of index of selected variables
- test_score1D array, (n_features, )
vector of test statistic
- thresfloat
knockoff threshold
- X_tilde2D array, (n_samples, n_features)
knockoff design matrix
References