Determination of the California Bearing Ratio of the Subgrade and Granular Base Using Artificial Neural Networks

The objective of the research is to estimate value California bearing ratio (CBR) through application ANN. methodology consists creating a database with soil index and CBR variables subgrades granular base pavements in Jaen, Peru, carried out mechanics laboratories city National University Jaen. In addition, Python library Seaborn for variable selection relevance, scikit-learn Keras libraries were used learning, training, validation stage. Five ANN are proposed value, obtaining an error 4.47% It can be concluded that this method effective valid determine bases any pavement its evaluation or design.