Graph inference datasets. Replication Data for: "Learning Functional Causal Models with Generative Neural Networks"

Graph datasets in csv format. Used in the article Learning Functional Causal Models with Generative Neural Networks. 1) Each file *_numdata.csv contain the data of around 20 variables connected in a graph without hidden variables. G2, G3, G4 and G5 refered to graph with 2, 3, 4 and 5 parents maximum for each node. Each file *_target.csv contains the ground truth of the graph with cause -> effect File beginning by "Big" are larger graphs with 100 variables. 2) Each file *_confounders_numdata.csv contain the data of around 20 variables connected in a graph. There are 3 hidden variables. Each file *_confounders_skeleton.csv contains the skeleton of the graph (including spurious links due to common hidden cause). Each file *_confounders_target.csv contains the ground truth of the graph with the direct visible cause -> effect. The task is to recover the direct visible links cause->effect while removing the spurious links of the skeleton