Figure 2From: Ciruvis: a web-based tool for rule networks and interaction detection using rule-based classifiersCorrelated features mask weak interactions. Rule networks for the outcome “0” in the simulated data. The data parameters are X = 0.10, Y = 0.15. (A) Using all features, (B) after the removal of the two strongest correlated features C4_15 and C3_11 and, (C) after the removal of the four strongest correlated features C1_4-C4_15. Connections between interacting features were colored black.Back to article page