If the data for the classes is imbalanced what are the best machine learning algorithms and will he choice vary in the case of binary and multi-class classification
In general, it is not possible to determine beforehand the best learning algorithm for a task (a corollary of no free-lunch theorem), as this depende on data and algorithm characteristics. There are some learning algorithm which may be less sensitive, but this is a general guide, and you must find an algorithm which fits you data well. Some treatment methods may help. See our paper:
Article Class imbalance revisited: A new experimental setup to asses...
Dear Muhammad! Please try to study the paper in the attachment (below). Vladimir
A classic technique is described in:
https://www.jair.org/media/953/live-953-2037-jair.pdf
In general, it is not possible to determine beforehand the best learning algorithm for a task (a corollary of no free-lunch theorem), as this depende on data and algorithm characteristics. There are some learning algorithm which may be less sensitive, but this is a general guide, and you must find an algorithm which fits you data well. Some treatment methods may help. See our paper:
Article Class imbalance revisited: A new experimental setup to asses...
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Learning