Single validation routines Sample Clauses

Single validation routines. Single validation routines judge the predictive performance achieved on a single test dataset. Several techniques are supported:  Training test set validation15 The user has to specify training and test dataset. A predictive model is build based on the training dataset, and applied to the test dataset. This is a common use case, as publicly available datasets are often already separated into training and test datasets.  Training test split validation16 The user has to specify a dataset. This dataset is split automatically into training and test dataset (The split is performed randomly; however, the random seed can be set by the user to repeat the exact same split.) A training test set validation is then performed with the two datasets.  Bootstrapping17 This is a state-of-the art mechanism that splits a single dataset into training and test dataset via sampling18. Again a training test set validation is performed with the created datasets.  Test set validation19 A test set validation is started with an already existing prediction model, applied to a test dataset. 14 This section provides an overview of existing validation methods. More details are provided in the API section as well as in previous Deliverable documents 5.1, 5.2 and 5.3. 15 xxxx://xxxxxxx.xxxxxxxxxx.xxx-xxxxxxxx.xx/validation/training_test_validation 18 xxxx://xx.xxxxxxxxx.xxx/wiki/Bootstrapping_%28machine_learning%29 19 xxxx://xxxxxxx.xxxxxxxxxx.xxx-xxxxxxxx.xx/validation/test_set_validation 3.1.1 Running a single validation example using HTML forms Figure 1: Use HTML form to invoke a training test split validation Figure 2: Use HTML form to build a report for the validation