Predictive analytics can be broken down into three broad categories: Recommender, Classification, Clustering
- Recommender—Recommender systems suggest items based on past behavior or interest. These items can be other users in a social network, or products and services in retail websites. There are some algorithm like Pearson correlation and euclidean distance.
- Classification—Classification (otherwise known as supervised learning) infers or assigns a category to previously unseen data, based on discoveries made from some prior observations about similar data. Examples of classification include email spam filtering and detection of fraudulent credit card transactions.
- Clustering—A clustering system (also known as unsupervised learning) groups data together into clusters. It does so without learning the characteristics about related data. Clustering is useful when you’re trying to discover hidden structures in your data, such as user habits.