There are various applications of clustering which are as follows −
Scalability − Some clustering algorithms work well in small data sets including less than 200 data objects; however, a huge database can include millions of objects. Clustering on a sample of a given huge data set can lead to biased results. There are highly scalable clustering algorithms are required.
Ability to deal with different types of attributes − Some algorithms are designed to cluster interval-based (numerical) records. However, applications can require clustering several types of data, including binary, categorical (nominal), and ordinal data, or a combination of these data types.
Discovery of clusters with arbitrary shape − Some clustering algorithms determine clusters depending on Euclidean or Manhattan distance measures. Algorithms based on such distance measures tend to discover spherical clusters with the same size and density. However, a cluster can be of any shape. It is essential to develop algorithms that can identify clusters of arbitrary shapes.
Minimal requirements for domain knowledge to determine input parameters − Some clustering algorithms needed users to input specific parameters in cluster analysis (including the number of desired clusters). The clustering results are quite sensitive to input parameters. Parameters are hard to decide, specifically for data sets including high-dimensional objects. This not only burdens users but also creates the quality of clustering tough to control.
Ability to deal with noisy data − Some real-world databases include outliers or missing, unknown, or erroneous records. Some clustering algorithms are sensitive to such data and may lead to clusters of poor quality.
Insensitivity to the order of input records − Some clustering algorithms are responsive to the order of input data, e.g., the similar set of data, when presented with multiple orderings to such an algorithm, and it can generate dramatically different clusters. It is essential to develop algorithms that are unresponsive to the order of input.
High dimensionality − A database or a data warehouse can include several dimensions or attributes. Some clustering algorithms are best at managing low- dimensional data, containing only two to three dimensions. Human eyes are best at determining the quality of clustering for up to three dimensions. It is disputing to cluster data objects in high-dimensional space, especially considering that data in high-dimensional space can be very inadequate and highly misrepresented.
Constraint-based clustering − Real-world applications can be required to perform clustering under several types of constraints. Consider that your job is to select the areas for a given number of new automatic cash stations (ATMs) in a city.