Validation
This section provides a description of current OpenTox validation components.
Validation Components
test-set validation
A prediction model and a test dataset are given. The model is applied to the test dataset. The prediction result is compared to the original values.
training-test-set validation
An algorithm, a training dataset and a test dataset are given. A model ist built on the training dataset and applied to the test dataset. The prediction result is compared to the original values.
bootstrap validation
An algorithm and a dataset is given. A training and and test dataset are produced via bootstrapping. A training-test-set validation is performed.
training-test-split validation
An algorithm and a dataset is given. A training and and test dataset are produced via splitting. A training-test-set validation is performed.
k-fold crossvalidation
An algorithm and a dataset is given. The dataset is split into k training and k test datasets. k training-test-set validation are performed.
lazar
lazar provides output including actual vs predicted values and validation statistics including a confidence index for every individual prediction.