Studying Tomek Link part 3:
When we shouldn't use Tomek Link:
1. Extremely Large datasets: coz takes high computational cost.
2. As it discards data points of two opposite classes which are closest. That makes inefficient for too much imbalance data.
#datascience
#imbalancedata
#tomeklink
Studying Tomek Link, day -2:
When to use Tomek Link:
- Dataset doesn't have extreme imbalance.
- Decision boundary is noisy.
- Preprocessing step for other techniques.
#tomeklink
#datascience
#imbalancedata
#undersampling
Studied Tomek Link: An undersampling technique, removes majority class samples, that are nearest neighbors of minority class samples.
#undersampling
#tomeklink
#datascience
#imbalancedata
#balancingdatasets
