Model Building

The critical step in QSAR modeling is the selection of appropriate modeling algorithm, the parameters for the algorithm and training the models. A wide range of algorithms have been employed by modelers, the choice of them depends largely on the data that is being modeled.

Sarchitect comes with the state-of-the-art in modeling algorithms. It has a large collection of algorithms for building classification and regression models.

For classification models

Naïve Bayes
Decision Trees
Neural Networks
SVM
Decision Forests

For regression models

Multiple Linear Regression
Neural Networks
SVM
Regression Forests
Partial Least Squares

Also, GA method that can use any of the above algorithms.

Nested and Ensemble model building is also supported.

The model building ‘Workflows’ in Sarchitect guide users through the model building steps. The results of each step are saved and captured on the ‘Navigator’ – this makes the iterative process of model building very easy.

Flowcharts capturing the approaches to model building suggested by our experts are available for modelers in the product.