Advanced Cross Validation Techniques in RapidMiner
You can read this in one minute. This post has been viewed 172 times.
My old colleague Tong Ji explains some advanced concepts of Cross Validation in RapidMiner.
My notes from the video are below:
- Another way to split the data in Batch Cross Validation (Batch CV)
- Batch CV you an specify certain values to split on
- Example: You run a sales offices in many regions, you can use Batch CV to split on the regions and use the other regions for testing.
- RapidMiner example uses Sonar data and splits on Region
- Sliding Window Validation operator is used in time series data
- The generic performance operator is the Performance operator
- Performance operator gives you ROC, Accuracy, Confusion Matrix)
- There are specific performance operators for Classification tasks, Regression Tasks, etc.
- Question: What are the difference/similarities with CV and fitness tests?
- Question: If you get different results from different validations, how do you interpret this?