Behavioral Predictive Framework


The elanyo EFPF Behavioral Predictive Framework (BPF) provides predictive insights for specific business use cases, initially only for the churn prediction use case. The framework has been built to potentially serve several use cases in one coherent place, allowing users to gain predictive insights by providing data resources upon which various models can be created that utilize the framework’s underlying algorithms. These models can then be used to score data, thus estimating the probabilities for different events, such as customer churn.

Following use cases are offered within the framework:

  • Churn Prediction

Introduction to Churn Predicition

One provided use case is the prediction of potential churners of a company. Machine learning algorithms analyze the probabilities of customers becoming potential churners and therefore no longer have a business relationship with the company. The algorithm achieves this by taking information about the customer (e.g. transaction data over the last 6 month, interactions with the company website, etc.) and by creating an instance of a model that is able to predict the outcome of a target variable which defines churned customers. This target variable is individually set based on the companies definition of a churned customer. The definition can vary from customers that stop buying the offered products to customers that stop interacting with the communication channels. The users of the BPF can later score their latest customer data in order to predict potential churners with the help of the model that has previously been created.

Current Deployment & Technologies

BPF minikube

Fig. 1 Deployments on Azure

Behavioral Predictive Framework Documentation

Documentation for Developers

Documentation for Users

Quickstart Guide