As we know, Vp and Vs are the main physical properties that can determine lithology from well log data or seismic data. But, in the real world the Vs data rarely found on the well log data, it is because the measurement cost is too pricey. So in this case, I used to generate and predict the value of Vs from the model. Of course to build the model, we need the neighboring well log data with the same well log parameter and has the Vs value. After, I run some of Machine Learning regression technique such as Decision Tree, Random Forest, XGBoost, and Artificial Neural Network. The result shows, the Artificial Neural Network algorithm outperformed the accuracy test and the Vs value generate from this model can be used to run prestack seismic inversion.
Skills: Python, Machine Learning, Deep Learning, Numpy, Lasio, Pandas, Matplotlib, Scikit-Learn