Demonstrates setting up a retraining experiment for programmatic update of the Telco Customer Churn model using new data.
This is the second part of the two part example based on the [Telco Customer Churn sample]. In the [first part], we created a predictive experiment to be published as a web service used for real-time and batch predictions. In this second part, we create a retraining experiment used to retrain the model programmatically with new data. ![enter image description here] After deploying this experiment as a web service, we can use the [C# sample code] to set up retraining in conjunction with the Predictive web service created in the first part. The retraining steps are explained in [this document]. At a high level, they are: 1. Add a new endpoint to the Predictive web service from first part (see sample code referenced above for more details) 2. Call the Retraining Web Service from this example (using the BES interface) 3. With the model trained in step 2, replace the model in the endpoint created in 1 (see sample code referenced above for more details) 4. Call the new endpoint, which will now be using the newly trained model, for predictions : https://gallery.cortanaintelligence.com/Experiment/Telco-Customer-Churn-5 : https://gallery.cortanaintelligence.com/Experiment/Telco-Customer-Churn-Predictive-Exp-1 : https://operationalizatio.blob.core.windows.net/publicimages/GalleryTelcoRetrain.PNG : https://github.com/raymondlaghaeian/TelcoCustomerChurn/blob/master/TelcoCustomerChurn/Program.cs : https://azure.microsoft.com/en-us/documentation/articles/machine-learning-retrain-models-programmatically/