Commercial use of
clustering
A retail grocery use clustering to segment its customer loyalty card into 1.3mm 5 different groups based on their buying behavior. It then adopted marketing strategies customized to each of these segments to target them more effectively.
One group was called the Fresh Food Lovers. These were customers who buy a large share of organic foods, fresh vegetables, salads, etc. A marketing campaign that emphasized the freshness of fruits and vegetables all year round availability of organic products in stores that appeal to client group.
The second group was called the Convenience junkies. These were people who buy frozen / semi-cooked, easy to prepare meals. The marketing campaign focuses on the internal bus speed of the distributor of frozen meals as well as banks store check-out worked well for this audience.
In this way, the retailer was able to deliver the right message to the right customer and maximize the effectiveness of its marketing.
Clustering Features
Clustering is a technique for extracting data undirected. That means it can be used to identify hidden patterns and data structures without having to formulate a specific hypothesis. There is no target variable in the clusters. In the above case was food retailer is not actively trying to identify those who love fresh produce at the beginning of the analysis. He was just trying to understand the purchasing behavior of its customers.
The grouping is performed to identify similarities in terms of specific behaviors or dimensions. In our example, the objective was to identify segments of customers with similar buying behavior. Therefore, clustering is performed using variables that represent the shopping habits of customers.
Cluster analysis can be used to discover data structures without an explanation or interpretation. In other words, cluster analysis simply discovers patterns in data without explaining why they exist. The resulting groups are meaningless by themselves. They have to be largely in the form of building their identity to know to understand what they represent and how they differ from the original population.
If the retailer was looming on the purchasing behavior of each pole. Customers of Group 1 spent a quarter of their total consumption of fresh and organic ingredients. This rate was considerably higher than other customers who spent less than 5% of this category. This customer segment called lovers of fresh food because that is what sets them apart from the rest of the clientele.
Types of
Clustering
There are several algorithms available to the group, and each can give another set of groups. The choice of a particular method depends on the goal of reunification, the desired output type, hardware and software services available and the size of the dataset. In general, clustering techniques can be divided into two categories according to the group structure they produce.
Non-hierarchical methods divide a set of data objects into groups N M. K-means, non-hierarchical technique, is the most used in the analysis of a company.
Hierarchical methods produce a series of nested groups in which each pair of objects found or groups in an increasingly bigger until only one group.
When should you use clustering?
The grouping is mainly used for
segmentation, as customers, product or store. Weve talked about customer segmentation using cluster analysis in the example above. Similarly, products can be divided into hierarchical groups based on their properties by the use, size, brand, flavor, etc. stores with similar characteristics like a turnover, size, clientele, etc. can be grouped.
Clustering can also be used for anomaly detection, for example, to identify fraudulent transactions. Cluster detection methods can be used on a sample containing only operations to determine the shape and size of the normal cluster. When a long transaction is not covered by the cluster for some reason, it is suspect. This approach has been used in medicine to detect abnormal cells in tissue samples and telecommunications to detect calling patterns, signs of fraud.
Clustering is often used to break large data set into smaller groups that are more susceptible to other techniques. For example, the results of logistic regression can be improved by running separately in smaller groups that behave differently and may have slightly different distributions.
In summary, clustering is a powerful technique for studying patterns in data structures and wide application is the analysis of
business. There are several methods for clustering. An analyst should be familiar with various clustering algorithms and should be able to apply the most appropriate technique depending on operational needs.
