In the field of forecasting the reshaped profile of slopes, Popov (1960) [13] studied stable slopes in the coastal zone by means of physical modeling. The stability of the remodeled profile in slope breakwaters with rock or concrete cube armor was studied by Priest et al. (1964) [14]. Van der Meer (1992) published the first version of his computer software, named "BREAKWAT [15]", to predict the reshaped profile of berm breakwaters. According to the studies carried out by Lykke Anderson (2006), the method of Van der Meer (1992) predicts the remodeled profile of dynamically stable berm breakwaters (H0T0>70) with acceptable accuracy. But for statically stable berm breakwaters (H0T0>70), the "BREAKWAT" software predicts overestimated damage for the breakwater. Furthermore, in this method, the cross-sectional areas of the eroded and accumulated parts are assumed to be equivalent, while according to the experimental results, due to the compression phenomenon of the material, these two areas are not necessarily equal. Ezabad et al. (2005) [16] presented a computer software called “IB” in which the reshaped profile of a rubble embankment structure could be estimated, using certain equations based on statistical models.2. Experimental dataA total of 412 test results are used in the present study, obtained from the experiments carried out by Moghim (2009) [17] and Shekari (2013) [18]. The ranges of these effective parameters covered by the tests are listed in Table 2. The material properties related to the different shielding and filter layers are listed in Table 3.3. Model treeModel tree is one of the machine learning approaches that makes the complex configuration of some modeling topics appear intractable by dividing them into simple...... middle of paper ..... The error for these estimates is given in Table 12.6. ConclusionsIn this study involving a total of 412 tests and implementing the M5 machine learning approach, seven new formulas for seven fundamental remodeling parameters are piloted. Using these seven equations, a geometric algorithm is written into a computer program that could predict the reshaped profile of the seaward slope of the berm breakwater. The following results can be drawn from the present study:- The predictive accuracy of model trees, although they construct linear regression models, was found to be sufficiently high in estimating parameter remodeling equations- According to Figures 7 to 14 and Table 11, the new program introduced in this study shows better results for the validation indices compared to those of BREAKWAT3 in the range of parameters governed by the existing datasets..