The aim of this contest was the development of data-driven models to estimate reservoir properties, including shale volume, porosity, and fluid saturation, based on a common set of well logs like gamma ray, bulk density, neutron porosity, resistivity, and sonic. Log data from eight wells and from the same field were provided, together with the corresponding reservoir properties estimated by petrophysicists. The goal was building a data-driven model using the provided training dataset. Following that, the newly developed data-driven models was deployed on the test data set in order to predict the reservoir properties based on the well-log data.
The notebook called Iron486_1.ipynb is the one with the best root-mean-squared-error on test dataset.
The file named report_Iron486_1.pdf is the report that includes the description of the methods and the obtained results.
In the folder called Other results there are other notebook results, using different hyperparameters and different techniques.
Iron486_1.csv is the file with all the predicted values on the test dataset.