From: Healthcare predictive analytics using machine learning and deep learning techniques: a survey
Method | Advantages | Disadvantages |
---|---|---|
K-means | • Easy-to-understand algorithm with linear relative computational complexity | • The user must specify the number of clusters or classes • Poor performance with irregular shape clusters |
Principal component analysis | • The use of PCA reduced the complexity of image grouping • Smaller database representation because only the trainee images are stored on a reduced basis in the form of their projections • Noise reduction because the maximum variation basis is used, so small variations in the background are ignored automatically | • It is difficult to evaluate the covariance matrix accurately • The PCA cannot capture even the most basic invariance unless the training data explicitly provide this information |
Apriori | • Makes use of the large itemset property • Simple to implement • It is simple to parallelize | • There are a large number of candidates sets generated • A number of database scans are required • The system I/O cost rises as a result of multiple scans of the transactional database |