Predicting brittleness indices of prospective shale formations from sparse well-log suites assisted by derivative and volatility attributes

A technique is proposed that calculates derivative and volatility attributes from just a few well log curves to assist in brittleness index predictions sparse well-log datasets with machine learning methods. Six are calculated for selected recorded logs: the first derivative, moving average of second logarithm instantaneous volatility, standard deviation volatility. These make it possible extrapolate calibrations cored comprehensively logged wells surrounding which only minimal suites recorded. Data two penetrating lower Barnett Shale distinct lithology five logs used demonstrate technique. Based on multi-K-fold cross validation analysis, data matching K-nearest neighbour model provides most accurate predictions, closely followed by tree-ensemble models. For this dataset, three plus matches prediction accuracy achieved logs. Moreover, any one their yields better performance than combination those Analysis Gini indices models reveals relative influences solutions. Such information perform feature selection optimize involved generate reliable predictions. Cited as: Wood, D. A. Predicting prospective shale formations assisted attributes. Advances Geo-Energy Research, 2022, 6(4): 334-346. https://doi.org/10.46690/ager.2022.04.08