Selecting a training policy

The siblings policy is used by default on the local classifier per node, but the remaining ones can be selected with the parameter binary_policy, for example:

• Exclusive
• Less exclusive
• Less inclusive
• Inclusive
• Siblings
• Exclusive siblings

rf = RandomForestClassifier()
classifier = LocalClassifierPerNode(local_classifier=rf, binary_policy="exclusive")

rf = RandomForestClassifier()
classifier = LocalClassifierPerNode(local_classifier=rf, binary_policy="less_exclusive")

rf = RandomForestClassifier()
classifier = LocalClassifierPerNode(local_classifier=rf, binary_policy="less_inclusive")

rf = RandomForestClassifier()
classifier = LocalClassifierPerNode(local_classifier=rf, binary_policy="inclusive")

rf = RandomForestClassifier()
classifier = LocalClassifierPerNode(local_classifier=rf, binary_policy="siblings")

rf = RandomForestClassifier()
classifier = LocalClassifierPerNode(local_classifier=rf, binary_policy="exclusive_siblings")