params |
Oject
|
Configuration of the training procedure
and decision tree architecture.
Properties
| Name |
Type |
Attributes |
Description |
min_impurity_decrease |
Number
|
<optional>
|
Minimal decrease
in criterion value (e.g. impurity) to form a split. Non - zero
split criterion could lead to problems on some of the datasets. |
min_samples_split |
Number
|
<optional>
|
Minimal number of
samples from the training dataset, with which a split can be
formed. If the value is in the range of [0.0, 1.0], then the
value is assumed to denote the fraction of the training dataset.
If the value is greater equal 2, this is assumed to be the total
number of samples. |
max_depth |
Number
|
<optional>
|
Maximum depth of the decision
tree allowed. |
|