Closed
Description
Description
In the paper of iforest the formula of average path length of unsuccessful search in BST as
avgPathLen = 2. * (np.log(n_samples_leaf - 1.) + 0.5772156649) - 2. * (
n_samples_leaf - 1.) / n_samples_leafbut in code sklearn.ensemble.iforest._average_path_length the formula is
2 * (np.log(n_samples_leaf ) + 0.5772156649) - 2. * (
n_samples_leaf - 1.) / n_samples_leafSteps/Code to Reproduce
Actual Results
if isinstance(n_samples_leaf, INTEGER_TYPES):
if n_samples_leaf <= 1:
return 1.
else:
return 2. * (np.log(n_samples_leaf) + 0.5772156649) - 2. * (
n_samples_leaf - 1.) / n_samples_leaf
average_path_length[not_mask] = 2. * (
np.log(n_samples_leaf[not_mask]) + 0.5772156649) - 2. * (
n_samples_leaf[not_mask] - 1.) / n_samples_leaf[not_mask]
-->
Expected Results
if isinstance(n_samples_leaf, INTEGER_TYPES):
if n_samples_leaf <= 1:
return 1.
else:
return 2. * (np.log(n_samples_leaf - 1.0) + 0.5772156649) - 2. * (
n_samples_leaf - 1.) / n_samples_leaf
average_path_length[not_mask] = 2. * (
np.log(n_samples_leaf[not_mask] - 1.) + 0.5772156649) - 2. * (
n_samples_leaf[not_mask] - 1.) / n_samples_leaf[not_mask]