softwareRisk (version 0.1.0)

path_fix_heatmap: Path-level improvement from fixing high-risk nodes

Description

Compute how much the risk score of the riskiest paths would decrease if selected high-risk nodes were made perfectly reliable (risk fixed to 0), and visualise the result as a heatmap.

Usage

path_fix_heatmap(all_paths_out, n_nodes = 20, k_paths = 20)

Value

A list with two elements:

• delta_tbl: a tibble with columns node, path_id and deltaR, containing the reduction in path risk score when fixing the node in that path.

• plot: a ggplot2 object containing the heatmap.

Arguments

all_paths_out

A list returned by all_paths_fun(), with elements nodes and paths. nodes must contain at least the columns name and risk_score. paths must contain at least the columns path_id, path_nodes (list-column of node names) and path_risk_score.

n_nodes

Integer, number of top-risk nodes (by risk_score) to include as rows in the heatmap. Defaults to 20.

k_paths

Integer, number of top-risk paths (by path_risk_score) to include as columns in the heatmap. Defaults to 20.

Details

For each of the top n_nodes nodes ranked by risk_score and the top k_paths paths ranked by path_risk_score, the function sets the risk of that node to 0 along the path (for all its occurrences) and recomputes the path risk score under the independence assumption, using

$$P_k = 1 - \prod_{i=1}^{n_k} (1 - r_{k(v_i)})$$

The improvement

$$\Delta P_k = R_{\mathrm{orig}} - R_{\mathrm{fix}}$$

is used as the fill value in the heatmap cells.

Bright cells indicate nodes that act as chokepoints for a given path. Rows with many bright cells correspond to nodes whose refactoring would improve many risky paths (global chokepoints), while columns with a few very bright cells correspond to paths dominated by a single risky node.

Examples

data(synthetic_graph)
out <- all_paths_fun(graph = synthetic_graph, alpha = 0.6, beta = 0.3,
gamma = 0.1, complexity_col = "cyclo")
res <- path_fix_heatmap(all_paths_out = out, n_nodes = 20, k_paths = 20)
res