What are the options for predictions in the customer churn prediction template?

This article details the options available for predictions in the customer churn prediction template.

Once models are trained in a customer churn prediction app, they can be used to make predictions. The template provides two options for making predictions:

• "One-off prediction"

• "Schedule periodic predictions”

The two options for running predictions in a customer churn prediction app: "one-off prediction" and "schedule periodic predictions".

Further explanation about the two options are provided below:

1. Make a one-off prediction: this option allows users to quickly generate future predictions of churn by:

   1. Either using the original input datasets, or

   2. By uploading new or updated data depending on the churn option selected when the app is built:

      1. For the transactional option: new transactions and updated customer-information data

      2. For the subscription option: new events of every event type defined while building the app, the updated subscription start and churn dates dataset, and the updated customer-information data

Options for one-off predictions

2. Schedule periodic predictions: This option allows users to get periodic predictions of churn probabilities for their current active customers, by simply providing live database connections to collect updated data of the types specified during the app creation (transaction, events of different kinds, subscription start and churn dates, or customer information).

Based on the user’s desired schedule, the Engine will collect the latest data from the provided connections, run predictions, and export the results to the user’s preferred destinations. This option is recommended for users who want to use a model in production where their datasets are regularly updated and new predictions are required on a regular basis.

Setting up input database connection

Defining output schedule

Defining output destination