Inspiration
All of the efforts that you put into obtaining a customer in the customer journey is useless if they don't stay on the platform. If we can find a way to predict if a customer is likely to leave your platform, the business can take preventative actions to assist the customer. Ex: talking to the customer to gain qualitative insights or even give them a promotional offer / incentive.
We give the business a signal to act on, before the customer potentially leaves.
What does it do
Given Customer data our model can predict whether a customer is likely to churn or not.
How we built it
We built the model using sklearn, xgboost and one hot encoding. We then packaged everything using a python front-end framework called stream-lit.
Challenges we ran into
We ran into an issue when calling the prediction model on our front-end framework. There was an issue connecting the front-end and the back-end together. We were stuck on it for a couple hours.
Accomplishments that we're proud of
Getting the model to run on the browser for the first time correctly was a very happy moment for us. We high-fived and said, "Let's go".
What we learned
This was our first time using the streamlit library and the result looked amazing.
What's next for Retained (Prediction model optimized for Customer Churn)
The next feature to add would be predicting when a customer is expected to churn. So giving a date.
Built With
- python
- xgboost
Log in or sign up for Devpost to join the conversation.