Classification is a well-known task in data mining, machine learning, and data science. This task aims to predict the class label for unseen instance as accurate as possible. In general, classification could be divided into two main types: Single Label Classification (SLC) and Multi-Label Classification (MLC). The main difference between SLC and MLC is that: MLC allows instances in the dataset to be associated with one or more class labels while SLC does not. This research is more interested in MLC. In specific, two main objectives are considered in this research. The first objective considers the identifying of the best base classifier that can handle the problem of MLC, while the second objective is the determining of the importance of discovering and exploiting the existing correlations among class labels in solving the problem of MLC. Extensive evaluations showed that RandomForest-X is the best classifier to handle MLC. Also, capturing high order correlations or ignoring these correlations could help in finding a solution for MLC problem. Moreover, the results revealed that Most Frequent Label (MFL) showed the best performance and utilizing the frequency of labels as a transformation criterion is better than utilizing the positive correlations and produces better results. Keywords: classification, machine learning, multi-label classification, multi-label ranking,
prediction, problem transformation methods.