Deploying an Analysis Model

  • Article

To deploy your analysis models to your Web servers, you can have your site developer modify the Global.asa file so that it contains a call to the LoadModelFromDB method. This call loads the Prediction model from your Commerce Server Data Warehouse onto your Web servers when the application is started. This method depends on your Data Warehouse always being available. If you cannot depend on your Data Warehouse to be available all the time, then use the following method.

You can deploy analysis models programmatically if you cannot ensure that the Data Warehouse you are loading your model from will always be available, or if you are concerned about minimizing your site re-initialization time. Be aware that this method does constitute extra work for your site developer.

You can have your site developer write a script that loads the model from the Data Warehouse (LoadModelFromDB), saves it to a file (SaveModelToFile) on the computer containing the Predictor resource, and then copies that file to each Web server.

After the model is built, the site developer must modify the ASP file located on each Web server in order to load the model from the file in which it was saved (LoadModelFromFile) onto each server. The site developer then must call one of the prediction methods (Predict, PredictMostLikelySegment, or PredictAllSegments) with the appropriate parameters.

Ee797522.note(en-US,CS.10).gif Note

  • Although your site developer can use the prediction methods to create both Segment and Prediction models, do not use Segment models for real-time recommendations because they are not optimized for this purpose and will not produce recommendations as accurate as Prediction models. Always build and use Prediction models for real-time recommendations on your site. You can also use the same model configuration to build a separate Segment model, based on the same data, and view it with the Segment Viewer in Commerce Server Business Desk. For information about using prediction methods, see Prediction Objects.

See Also

Deploying the Predictor Resource

Deploying Your Site


All rights reserved.