zenfiles/orbit-user-analysis

Managing your community with Orbit

Understanding the ZenML community has always been important aspect of our journey. It allows us to tailor our framework to meet the needs of our users, identify areas for improvement, build relationships with our community, and make data-driven decisions.

In order to achieve these goals, we use Orbit. Orbit is a community-building platform that provides a powerful set of tools to help you understand and manage your community better. From comprehensive community insights to engagement metrics, it helps us to track the growth of our tool and take the necessary steps going into the future.

Project description

In this project, we are using the Orbit API to detect members of our community who are either booming (highly active in the recent period) or churned(previously active, but recently inactive).

This is achieved by using a ZenML pipeline defined as community_analysis_pipeline which features three steps, namely booming, churned, and report:

• booming: detects members who conducted at least 150 events (default) in the last 14 days (default).
• churned: detects members who conducted at least one activity in the last 6 weeks (default) but no activity in the last 2 weeks (default).
• report: queries the current booming and churned users and saves them as output artifacts for tracking purposes. Moreover, it sends a message to a selected discord channel. The message includes a URL that directs you to an Orbit insight page which displays members from the last 6 months (default) with the related tags.

Installation

Install the required packages as follows:

pip install -r requirements.txt

Secrets

In order to run the pipelines, you need to have a stack with a secret manager and the following secrets

• orbit-api_token: the token to use the Orbit API
• orbit-workspace: the slug of the workspace in Orbit
• discord-webhook_url: the URL of the webhook that sends messages to a specific channel on Discord

Running it locally

After the installation is completed, you directly run the pipeline locally right away.

python run.py

Running it remotely

It is much more ideal to run a pipeline such as the community_analysis_pipeline on a regular schedule. In order to achieve that, you have to deploy ZenML and set up a stack that supports our scheduling feature. Please check our docs for more information.