Our portfolio

Have a look at our work

Customer churn modeling

The goal

To predict the probability of churn for the individual customer.

What we did

We developed a machine learning algorithm to achieve this goal. The project included the following steps:

  • Understanding the business problem and translating it to statistical problem
  • Analysis and evaluation of data availability and quality
  • Data preparation and transformation
  • Testing and tuning different predictive analytics algorithms (Boosting trees, Neural networks, Logistic Regression, Random Forest, etc.)

We also helped the company to deploy the best model and made recommendations on how the results can be used. Currently, the company is using our model for churn prediction and prevention.

  • Client: Telecom, Armenia