In this study, four CONCRETE types, including ordinary Portland cement CONCRETE, fly ash CONCRETE, slag CONCRETE, and slag-fly ash CONCRETE, are taken into account in order to estimate their compressive strength by two novel machine learning methods (genetic algorithm and soccer league competition algorithm), and four types of regressions (linear, 2nd order polynomial, exponential, and logarithmic). Subsequently, the precision of prediction models are compared based on performance indicators, and the most accurate models are applied in the optimization problem modeling. Drawing on results, the most precise model to estimate the compressive strength of ordinary Portland cement CONCRETE is the genetic algorithm, and the soccer league competition is the most accurate model to estimate the strength of other CONCRETE types. Afterward, a model is developed so as to design mixture proportions of 40MPa CONCRETEs. Fly ash CONCRETE, slag-fly ash CONCRETE, and slag CONCRETE reduce the unit cost by 35. 2%, 29. 9%, and 23. 1%, respectively, compared with ordinary Portland cement CONCRETE. Fly ash CONCRETE, slag-fly ash CONCRETE, slag CONCRETE, and ordinary Portland cement CONCRETE require 217. 25 kg, 150. 47 kg, 102 kg, and 414. 64 kg cement to be manufactured. Furthermore, the slag CONCRETE can reduce the amount of cement in the mixture proportion by 75. 4%, and it is the most eco-friendly CONCRETE.