Adaptive boosting of random forest algorithm for automatic petrophysical interpretation of well logs

The power of Machine Learning is demonstrated for automatic interpretation well logs and determining reservoir properties volume shale, porosity, water saturation respectively tight clastic sequences. Random Forest algorithms are reputed their efficiency as they belong to a class called ensemble methods, which traditionally seen weak learners, but can be transformed into strong performers promise deliver highly accurate results. study area located offshore Australia in the Poseidon Crown fields situated Browse Basin, gas complex reservoirs. There 5 wells used this with one manually interpreted subsequently developing machine learning model predicts output other 4 wells. basic open hole namely Natural gamma ray, Resistivity, Neutron Porosity, Bulk Density, P-wave S-wave sonic travel-time, interpretation. One has missing travel-time log was also predicted by model. results indicate very robust improvement performance when algorithm combined Adaptive Boosting interpreting logs. training accuracy using alone 98.21%, testing 77.62% suggested over-fitting resulted overall 99.40% an 97.03%, indicating drastic performance. preparing set consisting gave 99.79% 98.54%. successfully applied data sedimentary environment drastically improved Boosting.