Shear Strength analysis in beam-column joint is important to avoid failure and measure the performance of reinforced concrete in structures. This analysis is important because failures in the column beam joints can cause partial damage or collapse of the structures. Two types of failures are analyzed: brittle failure, which is by shear before the creep of the beam, and ductile failure by shear after the creep of the beam, the creep of the beam without joint failure. In this article, 598 rows \(\times \) 14 columns were studied, which are the input and one output characteristics (Shear Strength), 80% of the data was partitioned in training and 20% for testing. Supervised learning models (Decision Tree, Random Forest, Adaboost, GBRT and XGBoost) and Artificial Neural Networks and MLP were applied in the prediction of Shear Strength. In the case of supervised learning techniques, the metrics of MAE, MSE, RMSE, \(R^2\) , SMAPE were used; of which XGBoost gave better results in MAE = 0.3920, MSE = 0.3382, RMSE = 0.5816 and Random Forest in \(R^2 = 0.8138\) , XGBoost SMAPE = 0.4017; on the other hand, applying the Neural Networks SMAPE = 0.118, Acurracy = 72.12% and MLP with SMAPE = 0.116, Acurracy = 73.82%. One of the best models The Neural Networks with 500 neurons because it gave a better fit in SMAPE and a larger Acurracy.