In order to increase the efficiency of predicting concrete compressive strength, ensemble machine learning (ML) algorithms are required. Considering that each ML algorithm continuously varies in methodology, one ML algorithm cannot generate exhaustive prediction results since limited parameters are available to tune. This research serves as a beginning step towards predicting the compressive strength of concrete containing waste tyre rubber (WTR) and clay brick powder (CBP) using artificial neural network (ANN), random forest (RF), decision tree (DT) and support vector machine (SVM) algorithms. Taylor diagram model analysis shows that when the four algorithms are compared, the SVM (train) model demonstrates the highest performance in predicting the compressive strength of concrete containing CBP and WTR. The R2 values ranging from 0.60 – 0.97 imply that all the models fairly predict the compressive strength of concrete containing CBP and WTR. The same predictive abilities are demonstrated by the clustering of the data points for train and test models around the y = x line. It is shown that the majority of the data points lie within the error lines range of −20 and + 20 %. The SHapley Additive exPlanations (SHAP) analysis reveals that WTR has the highest impact on model predictions with a mean SHAP value of 3.83, while cement shows a moderate influence with a mean SHAP value of 0.77. Moreover, these findings suggest that WTR content is the most critical factor in controlling the concrete's compressive strength, while cement content plays a supporting role in the mixture design. Since the prediction behaviour of concrete using ML models is governed by the replacement levels of CBP and WTR, the models used in this study can be extended to the concrete mixes containing other waste materials.