Estimation of geomechanical parameters of rocks plays an important role in designing mining and civil projects. In general, the determination of these properties is done in two ways, directly and indirectly. Direct methods include laboratory tests, which, despite the high precision, are costly and time consuming. In addition, the preparation of core samples is difficult to perform for laboratory experiments, especially in soft rocks, and it requires very careful attention. The characteristics of soft rocks are unresolved issues among geotechnical materials, which have not been addressed to them like soils and hard rocks so far. Because of the intermediate properties of these materials, it is necessary to estimate their parameters with an appropriate approach. To overcome these problems, indirect methods can be used as a convenient and reliable alternative. Using indirect methods, two characteristics of uniaxial compressive strength (UCS) and elastic modulus (E) can be obtained employing the input variables including density, porosity, water content, water absorption and compressive velocity waves. In order to estimate UCS and E, predictive techniques such as artificial neural network, decision tree and linear regression and nonlinear regression techniques have been used in this study. In this research, after collecting geomechanical properties of soft rocks (such as siltstone, mudstone, marl, and claystone) resulting from laboratory experiments, the relationship between the UCS and E were set based on non-destructive tests including sound velocity and physical. The RMSE and R2 statistics were used to evaluate the resulting models. Different combination of the independent input variables were tried and the accuracy of the methods were studied. The results showed that the decision tree method was the most suitable method for establishing the relationship between parameters from destructive tests and parameters from non-destructive experiments