Customer churn is a significant problem for telecommunication service providers. Annual churn rates for telecommunications companies’ average between 10% and 67% globally. According to the 2016 IRJET report, the USA alone witnesses a 29% churn rate. As most of these customers originate from another wireless provider, they are already churners.
When a customer leaves, service providers lose not only future revenue but also the resources spent to acquire the customer in the first place. Such acquisition costs hundreds of dollars in the telecom industry. The ability to predict customer churn enables service providers to both retain and increase their customer base and consequently, their revenue. Churn prediction models built using Machine Learning techniques provide this ability.
Machine Learning is a term used to refer to software that mimics the human ability to extract knowledge from experience. Broadly speaking, Machine Learning algorithms identify patterns in historical data and then correlate these with events of interest, such as churn. These algorithms build a model which can then be deployed to make predictions.
The framework supports construction of appropriate data sets. To reach a particular goal, they must be trained using prepared data sets. The results are then presented to people or applications that can provide feedback, which leads to further training to refine these insights.
Components within the layered architecture pattern are organized into horizontal layers, each layer performs a specific role within the application (e.g., presentation logic or business logic). Each layer of this layered architecture pattern has a specific role and responsibility within the application. For instance, the presentation layer contains information on each individual customer’s behaviour, their data usage, applications used, calls made, preferred communications mode, birthdays, age, etc.
For example, a presentation layer would be responsible for handling all user interface and browser communication logic, whereas a business layer would be responsible for executing specific business rules associated with the request. Each layer in the architecture forms an abstraction around the work that needs to be done to satisfy a particular business request.
mViva, a Precision Marketing solution from Pelatro includes the capability to create and deploy these models. The user can choose the target variable (event of interest). An appropriate algorithm can also be selected, depending on the nature of the problem, and that of the available data. Datasets (features) can also be defined in real-time, and appropriate pre-processing can be specified – including Principal Component Analysis and outlier elimination.