Regression: Demand estimation - [Predictive Exp.]

May 30, 2016

568 views
141 downloads


Report Abuse
This experiment demonstrates setting up a predictive experiment for demand estimation and the steps to complete the retraining of its model.
This experiment builds on the [Regression: Demand estimation example][1] using a two-part sample. In the [first part][2], we created a retraining experiment. In this part, we create a predictive experiment which can be deployed as a web service (operationalization). ![Predictive Experiment][3] After deploying a web service from this experiment, we can use [C# code][4] to prepare for and complete the retraining. In that code, we: 1. Add a new scoring endpoint to the deployed scoring endpoint 2. Call the retraining web service to get a newly trained model 3. Update the new scoring endpoint with the new model After the completion of these steps, we will have a new endpoint with a retrained model. [1]: https://gallery.cortanaintelligence.com/Experiment/Regression-Demand-estimation-4 [2]: https://gallery.cortanaintelligence.com/Experiment/Regression-Demand-estimation-Retraining-Exp-1 [3]: https://operationalizatio.blob.core.windows.net/publicimages/BikingSampleGalleryExperiment2.PNG [4]: https://github.com/raymondlaghaeian/BikeRentalDemandEstimation