Abstarct: This study aims to use a screening technique to reduce the dimension of the data matrix. For this purpose, the new penalized Ridge preprocessing technique is used to reduce the data dimensions. Then, the Smoothly Clipped Absolute Deviation (SCAD) as a penalized variable selection method is used to select the most effective descxriptors and modeling. Selected descxriptors were used as input to construct the QSAR model using the SCAD method. For this purpose, the QSAR model was developed using multiple Ridge implementation as a preprocessing technique for data screening. The remained descxriptors were entered into SCAD penalized model. After implementing SCAD to the train set data, some effective descxriptors (9) were used in the model construction. The proposed optimum model (Ridge-SCAD) was evaluated using calculation of statistical parameter, applicability domain, and Y- Randomization test. The mentioned different tests show satisfactory results. The performance of the Ridge-SCAD model was applied for the suggestion of some potent c-Met inhibitors. The efficiency of the recommended compounds was investigated using molecular docking and calculation of Lipinski’s rule of five. The obtained results prove the potency of the new compounds. For the comparision aim, the QSAR model was developed without using the Ridge method as a preprocessing technique, and all the calculated descxriptors were defined as SCAD inputs. The SCAD model was constructed using the train set data, and 12 descxriptors were obtained as significant descxriptors. The biological activity of the test data was predicted using the optimum SCAD model. The statistical parameters of SCAD model were calculated and compared with the Ridge-SCAD model. The results proved that the high predictability of the Ridge-SCAD model against to the SCAD model.