Abstract <p>Estimating with low uncertainty the parameters of time and magnitude of upcoming earthquakes is necessary to create an earthquake warning system. Nowadays, by using different satellite data, it is possible to monitor a large number of earthquake precursors. Multi-precursor analysis, along with multi-method analysis, has made it possible to detect a large number of LAI (lithospheric atmospheric ionospheric) seismic anomalies in the study of strong earthquake-affected areas. In this study, the deviation values of 898 LAI anomalies detected using 20 implemented predictor algorithms around the time and location of 21 powerful earthquakes that occurred in recent years have been considered. Using different scenarios, various functions were fitted on the collected data, including the day of anomaly observation, anomaly intensity, geographic latitude of epicenter and real magnitude of the earthquake, and functions were developed to estimate magnitude parameters with RMSE of about 0.53 (M_W) and the day of the earthquake with about RMSE of 8.27 day. In addition, by using an MLP neural network, and training it using the detected LAI anomalies, accuracies of 0.21 and 9.29 were obtained, respectively, for estimating the magnitude and time of an impending earthquake. Therefore, by comparing the two functional and machine learning-based methods proposed in this study, it can be concluded that the proposed functions are efficient for estimating magnitude and time of forthcoming strong earthquakes. Although the accuracy of predicting the magnitude of the earthquake is acceptable, the accuracy of about 8 days for predicting the day of the earthquake can be efficient for relatively short-time earthquake prediction.</p>