The telecommunications operator Interconnect offers mainly two types of services, the first is landline communication where the telephone connection can be made to several lines simultaneously, and the second is the Internet service, where the connection to the network can be established via a telephone line (DSL, digital subscriber line) or via a fibre optic cable. In addition, users can benefit from other advantages such as online security, file storage in the cloud, data backup, technical hotline and TV and movie streaming. Moreover, the customer can choose between a monthly payment or sign a 1 or 2 year contract. They can also choose between several payment methods and get an electronic invoice after each transaction.
The company wants to be able to predict their cancellation rate. In case it detects that a user is planning to leave, they will be offered promotional codes and special plan options. Interconnect's marketing team has been tasked with collecting some of its customers' personal data, including information about their plans and contracts.
- Perform an initial analysis of the data and then process the data until no missing values, duplicates and other errors are recorded.
- Examine and discover patterns in the data through exploratory univariate and bivariate analyses.
- Classify the data and prepare machine learning models by coding, normalisation and other methods deemed relevant.
- Develop, refine and test various machine learning models in order to better classify users according to the likelihood that they will or will not cancel their Interconnect accounts.
- Choose a model that obtains an AUC-ROC greater than 0.88 on the test set.