Compute the number of times each edge was selected when performing a non-parametric bootstrap (see Figure 6.7, Hastie et al. 2009) .
boot_eip(Y, method = "pearson", samples = 500, progress = TRUE, ...)
Y | A matrix of dimensions n by p. |
---|---|
method | Character string. Which correlation coefficient (or covariance) is to be computed. One of "pearson" (default), "kendall", or "spearman". |
samples | Numeric. How many bootstrap samples (defaults to |
progress | Logical. Should a progress bar be included (defaults to |
... | Additional arguments passed to |
An object of class eip
that includes the "probabilities" in a
data frame.
Although Hastie et al. (2009) suggests this approach provides probabilities, to avoid confusion with Bayesian inference, these are better thought of as "probabilities" (or better yet proportions).
Hastie T, Tibshirani R, Friedman J (2009). The elements of statistical learning: data mining, inference, and prediction. Springer Science \& Business Media.
# \donttest{ # data (ptsd symptoms) Y <- GGMncv::ptsd[,1:10] # compute eip's boot_samps <- boot_eip(Y, samples = 100, progress = FALSE) boot_samps #> Edge Inclusion 'Probabilities': #> #> Relation EIP #> B1--B2 0.98 #> B1--B3 0.29 #> B2--B3 1.00 #> B1--B4 1.00 #> B2--B4 0.20 #> B3--B4 0.98 #> B1--B5 0.79 #> B2--B5 0.24 #> B3--B5 0.88 #> B4--B5 0.99 #> B1--C1 0.14 #> B2--C1 0.04 #> B3--C1 0.04 #> B4--C1 0.46 #> B5--C1 0.88 #> B1--C2 0.08 #> B2--C2 0.73 #> B3--C2 0.14 #> B4--C2 0.56 #> B5--C2 0.91 #> C1--C2 0.94 #> B1--D1 0.34 #> B2--D1 0.14 #> B3--D1 0.13 #> B4--D1 0.07 #> B5--D1 0.12 #> C1--D1 0.44 #> C2--D1 0.38 #> B1--D2 0.57 #> B2--D2 0.16 #> B3--D2 0.07 #> B4--D2 0.37 #> B5--D2 0.08 #> C1--D2 0.43 #> C2--D2 0.44 #> D1--D2 0.54 #> B1--D3 0.22 #> B2--D3 0.39 #> B3--D3 0.26 #> B4--D3 0.85 #> B5--D3 0.40 #> C1--D3 0.26 #> C2--D3 0.12 #> D1--D3 0.65 #> D2--D3 0.95 #> ----- # }