Description
This ratio specifies how examples are sampled from the majority class when the task is binary classification.
This means that for every example in the minority class, we sample majority_sampling_ratio
number of examples from the majority class. majority_sampling_ratio
must be > 0
in case it is specified.
For example, a ratio of 10 means that we will sample the examples of the majority class with ratio 1:10, or 10 majority class examples for every minority class example.
If the ratio of majority:minority
is greater than majority_sampling_ratio:1
, we will sample uniformly, ignoring this value.
For example, if the ratio of majority:minority = 10:1
and we choose majority_sampling_ratio=9
then we will ignore this setting.
Supported Task Types
- Binary Classification
Default Values
run_mode | Default Value |
---|---|
FAST | None |
NORMAL | None |
Updated 5 months ago