Model Builder

Select model template, training set and descriptor-coefficient-sheet

Create a model

Select the training and validation sets, the machine learning method and the validation protocol


Select the training and validation sets:


Training set (required): [...]
Add a validation set

Choose the learning method:
Suggested modeling methods:

Methods under development:

Model validation
Validation method:
Number of folds:
Stratified cross-validation (classification only)
Treat each record as a new molecule
Number of bagging models:
Stratified bagging (classification only)
Use custom number of instances in training set for each bag
Treat each record as a new molecule
Development of a model without validation may lead to incorrect estimation of the prediction accuracy of your model and incorrect assessment of its Applicability Domain.
Each member of the consensus model has its own validation method.

You can create a model from template: import an XML model template or use another model as a template