Why metrics : tpr != recall

[cosetteke]

i am using this package mt = MetricsDAG(est_causal_matrix, true_causal_matrix)

the source codedef _count_accuracy(B_est, B_true, decimal_num=4)
shows tpr: TP/(TP + FN)
recall: TP/(TP + FN)

why the result is different?

[cosetteke]

B_est
[[0 0 1 0 0 0 0 0 0 0 0 0 0 1 0 1 0 0]
[0 0 1 1 0 0 0 0 0 0 0 0 0 0 0 1 0 0]
[0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 1 0 0]
[0 0 1 0 0 1 0 0 0 0 0 0 0 0 0 1 0 0]
[0 0 0 0 0 1 1 1 0 0 0 0 0 0 0 0 1 0]
[0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0]
[0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 1 0 0]
[0 0 0 1 0 0 0 0 0 0 0 0 0 0 1 0 0 0]
[0 0 0 0 0 0 0 0 0 0 1 1 0 0 0 0 0 0]
[0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0]
[0 0 0 0 0 0 0 0 1 1 0 1 0 0 0 0 0 0]
[0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
[0 0 0 0 0 1 0 1 0 0 0 0 0 0 0 1 0 0]
[0 0 0 0 0 0 0 0 0 1 0 0 0 0 1 0 0 0]
[0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0]
[0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
[0 0 1 0 0 0 0 1 0 0 0 0 0 0 0 1 0 1]
[0 0 0 0 1 0 0 0 0 0 0 0 0 1 0 0 0 0]]
B_true
[[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1]
[1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 0 1]
[1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 0 1]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1]
[1 1 1 1 1 1 1 1 0 0 0 1 1 1 1 1 1 1]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0]
[1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1]]

[cosetteke]

{'fdr': 0.2703, 'tpr': 0.9739, 'fpr': 10.0, 'shd': 126, 'nnz': 37, 'precision': 0.7368, 'recall': 0.1045, 'F1': 0.183, 'gscore': 0.0672}

[shaido987]

Hello, thanks for your bug report. I believe I have identified the issue, you can see the related PR. In summary, the recall value should be correct while the tpr currently is incorrect due to how it is computed. So you can simply use the recall value for now or apply the patch locally first (it only requires changing two lines in the source code).