The reduction in compressive strength (CS) of cementitious composites incorporating waste plastic is the main concern limiting its applicability in the building sector. Using industrial wastes as cement substitutes to enhance the CS of plastic mortar is a sustainable approach. This study used fine powdered waste materials such as silica fume (SF), marble powder (MP), and glass powder (GP) in plastic-based mortar for their effectiveness in enhancing CS. Plastic mortar specimens were cast using plastic waste in 5–25 % contents as sand replacement by mass, and their 28-day CS was recorded as a reference. SF, GP, and MP were utilized in plastic mortar mixtures separately in proportions of 5–25 %, with a 5 % increment, substituting cement by mass. These waste powders were also used in combinations of two (SF+GP, SF+MP, and GP+MP) and three (SF+GP+MP) in plastic mortar mixtures. Moreover, prediction models were built using the experimental database for the CS of plastic mortar. Gradient boosting and bagging ensemble machine learning (ML) techniques were chosen for model development. The decrease in CS was limited by substituting SF, GP, and MP for cement in plastic mortar. It was determined that the most effective substitution levels for SF, GP, and MP in plastic mortar, according to the strength enhancement, were 15, 10, and 15 wt.% of cement, respectively. The ML models closely matched experimental results, and in terms of R2 and error evaluations, bagging model outputs were more accurate than gradient boosting. The gradient boosting and bagging models had R2 of 0.89 and 0.94, respectively, with average absolute errors of 0.87 and 0.65 MPa.