Cloud Pak for Data

 View Only

Cloud Pak for Data accelerators in wealth management | Customer  churn prediction

By Carson Masterson posted Tue July 09, 2019 06:50 PM

  

The following videos show how you can use the Customer Churn Prediction accelerator in IBM Cloud Pak for Data to identify customers who have are likely to churn.

Nicole, a data scientist, is tasked with identifying the customers who are likely to take their business elsewhere so that her company can reduce the rate of customer churn. She turns to the Customer Churn Prediction Accelerator to jump-start her analysis.
Customer churn prediction - Overview video  

This video provides a brief overview of the accelerator and highlights its usefulness for the wealth management industry.  Understand how the accelerator can use specific signals to detect client dissatisfaction and generate a churn risk score. 
Customer churn prediction – Data catalog video 

Before she can create a Customer Churn Prediction model, Nicole, the data scientist, needs to have access to the data. This video shows how a data steward can import the terms from the business glossary that is included in the Customer Churn Prediction accelerator. When the data steward runs automated discovery on the data sources that are connected to Cloud Pak for Data, the data is automatically mapped to the terms in the glossary. This makes it easier for Nicole to identify the right data to use in her analysis. 
Customer churn prediction – Data prep & ML model training video 
This video shows how Nicole, the data scientist, can use a sample Jupyter notebook to prepare her data and build and train a machine learning model that can predict customer churn. The notebook includes information about the script that she can run to cleanse and prepare her data. The notebook also helps her test three separate models to determine which model returns the most accurate predictions. 
Customer churn prediction – ML model scoring video 
Now that Nicole, the data scientist, has trained her machine learning model, she's ready to test and score the customer churn prediction model. This video shows how she uses a sample Jupyter Notebook that includes information about installing the prerequisites for the model, walks her through the process of testing and scoring the customer churn prediction model, and shows her how to deploy the model as a web service. 
Customer churn prediction – User API interface video 

This video shows how Nicole, the data scientist, can use the deployed web service and the sample dashboard in the accelerator to leverage the insight from the customer churn prediction model. The dashboard gives her a quick and easy way to identify which customers are most likely to churn and the factors that influenced the rating. The dashboard can help the company get to the root of the problem. 

#CloudPakforDataGroup
0 comments
36 views

Permalink