huge (version 0.7)

huge.roc: Draw ROC Curve for a solution path

Description

Draws ROC curve for a solution path according to the true graph structure

Usage

huge.roc(est, theta, ind.group, verbose = TRUE)

Arguments

est

There are two choices for the input est: (1) A solution path. (2)An object with S3 class "huge" containing est$path as a solution path, est$theta as the true graph structure and est$ind.group containing indices from

theta

The true graph structure. Only applicable when the input est is a solution path.

ind.group

A k dimensional vector containing k indices from a subset of all variables. Only applicable when estimating a subgraph of the whole graph and the input est is a solution path. The default value is c(1:d)

verbose

If verbose = FALSE, tracing information printing is disabled. The default value is TRUE.

Value

• An object with S3 class "roc" is returned:
• F1
  The F1 scores along the solution path.
• tp
  The true positive rates along the solution path
• fp
  The false positive rates along the solution paths
• AUC
  Area under the ROC curve

Details

For a lasso regression, the number of nonzero coefficients is at most n-1. If the dimension d>>n, even when lambda is small, the estimated graph may still be sparse. In this case, the AUC may not be appropriate to evaluate the performance.

References

Tuo Zhao and Han Liu. HUGE: A Package for High-dimensional Undirected Graph Estimation. Technical Report, Carnegie Mellon University, 2010

See Also

huge and huge-package

Examples

#generate data
L = huge.generator(d = 200)

#subset indices
ind.group = c(1:80)

#subgraph estimations using lasso
out1 = huge(L, ind.group = ind.group)

#draw ROC curve
Z1 = huge.roc(out1)

#Compare the maximum F1 score
max(Z1$F1)