You are viewing the RapidMiner Go documentation for version 9.5 - Check here for latest version
RapidMiner Go is being discontinued. For an alternative, please contact sales about our new AutoAI cloud solution.
By now, you have:
and you're ready to make predictions. The data you used to build your model represents the past. If it's an accurate description of the past, and the future data follows the same patterns, then your predictive model should be a success!
It's a fundamental assumption of predictive modeling that the future is not so different from the past. Don't forget, however, that it is an assumption, and it might or might not be true. If it is true, you will see that your errors in prediction are about the same for the future data as they were for the test set. That's a sign that you created a good model -- it works, at least for now!
If the predictions are worse than they were during training and testing, you should contemplate rebuilding your model to reflect the actual circumstances.
- Maybe the original data set did not accurately describe the situation.
- Maybe circumstances have changed; the future is not the same as the past.
Once you have created your model, RapidMiner Go provides two mechanisms for making predictions:
- Apply your model: upload a new data set and see the predictions
- Deploy your model: make your model available to other people and software
Of these, the first is more straightforward and the second is more flexible. To Apply your model, simply upload a new data set, compatible with the data you used to build your model, and the predictions will be displayed. Alternatively, if you want to write a program that uses RapidMiner Go to generate predictions, learn how to Deploy your model.