Case study – customer churn with PostgresML

Let’s revisit the customer churn problem for a telecommunications provider. As a reminder, the dataset consists of customers and their account and cost information associated with the telecommunication provider.

Data loading and preprocessing

Our data will typically already be available within the PostgreSQL database in a real-world setting. However, for our example, we will start by loading the data. First, we must create the table the data is loaded into:

CREATE TABLE pgml.telco_churn
(
    customerid       VARCHAR(100),
    gender           VARCHAR(100),
    seniorcitizen    BOOLEAN,
    partner          VARCHAR(10),
    dependents     ...