The total organic carbon content (TOC) is a core indicator for shale gas reservoir evaluations. Machine learning-based models can quickly and accurately predict TOC, which is of great significance for the production of shale gas. Based on conventional logs, the measured TOC values, and other data of 9 typical wells in the Jiaoshiba area of the Sichuan Basin, this paper performed a Bayesian linear regression and applied a random forest machine learning model to predict TOC values of the shale from the Wufeng Formation and the lower part of the Longmaxi Formation. The results showed that the TOC value prediction accuracy was improved by more than 50% by using the well-trained machine learning models compared with the traditional method in an overmature and tight shale. Using the halving random search cross-validation method to optimize hyperparameters can greatly improve the speed of building the model. Furthermore, excluding the factors that affect the log value other than the TOC and taking the corrected data as input data for training could improve the prediction accuracy of the random forest model by approximately 5%. Data can be easily updated with machine learning models, which is of primary importance for improving the efficiency of shale gas exploration and development.

The Linnan Sag is one of the main oil-producing units in the Huimin Depression, Eastern China, and the pore pressure gradients obtained from drill stem tests (DSTs) range from 9.0 to 16.0 MPa/km. Uncertainty about the origin and distribution of abnormally high pressures in the Linnan Sag has led to different interpretations of hydrocarbon accumulation and resource assessments, and it interferes with safe drilling. In the Linnan Sag, mudstone compaction curves are substantially affected by several non-compaction factors, and the normal trend of the compaction curve is difficult to determine. The determination of the origin and distribution of overpressure in the Linnan Sag is a challenge. In this study, the factors that may affect mudstone compaction—such as the shale volume, higher calcareous, and organic matter content—were carefully examined and processed. The pressures in the mudstones were estimated by the corrected mudstone compaction curves, which were compiled from acoustic, density, and neutron logs, and calibrated using DST and mud weight data. The log response–vertical effective stress and acoustic velocity-density crossplots were used to identify the mechanisms that generate overpressure. The comprehensive compaction curve shows that the mudstones in the overpressured layer exhibit clear disequilibrium compaction characteristics. The logging response crossplots demonstrate that those overpressured points were consistent with the loading curve. The findings suggest that, the fundamental mechanism resulting in overpressures is the disequilibrium compaction of thick Paleocene mudstones. Hydrocarbon generation and vertical transfer of overpressure may be the main unloading mechanisms, which corresponds to the overpressure points that deviate from the loading curves. Since organic matter cracking may occur in formations at depths greater than 4000 m (Ro > 1.0%), the contribution of hydrocarbon generation to overpressuring is expected to be limited. The transfer of overpressure through opening faults is therefore considered to be the main cause of higher overpressure in local sandstones. The overpressures in the mudstones are characterized by a gradual decrease from the center to the margin in the Linnan Sag. The pressure in the isolated sand bodies are generally similar to that in the surrounding mudstones, whereas it can be lower or higher when the overpressure in the sand bodies are vertically transferred by faults to other pressure systems. The results of this analysis provide an indication of the magnitude, mechanism, and distribution of overpressure in the Linnan Sag. This insight can be used to guide further exploration of the Linnan Sag and similar geological basins.