Processed Field-Experience Dataset for Explainable Machine-Learning EOR-Family Screening
Description
This dataset contains the processed analytical data used in the study “Explainable Machine Learning with Physics-Consistency Auditing for Field-Experience-Informed EOR Screening: A Niger Delta Sandstone Application.” It comprises 36 source-verified Grade-A historical enhanced oil recovery (EOR) field/project/reservoir observations used for primary machine-learning analysis: 11 Chemical, 12 Gas, and 13 Thermal cases. The six-variable feature set consists of porosity, permeability, reservoir depth, reservoir temperature, American Petroleum Institute (API) gravity, and oil viscosity. Missing values were not statistically imputed. Permeability and viscosity were log10-transformed for modelling. Historical EOR implementation is used as the classification target and should not be interpreted as universally optimal EOR selection. Niger Delta application cases, including ND04/X-Field, are excluded from model training. Case-level source provenance is documented separately in Supplementary Table S1 accompanying the associated manuscript.
Files
Steps to reproduce
The dataset was compiled from documented field enhanced oil recovery (EOR) projects and screened using source-level quality control. Each observation represents a distinct field/project/reservoir condition and is classified into Chemical, Gas, or Thermal EOR families according to the historically implemented process. The primary analytical dataset contains 36 Grade-A cases complete in six predictors: porosity, permeability, reservoir depth, reservoir temperature, API gravity, and oil viscosity. Missing values were not statistically imputed. Permeability and oil viscosity were log10-transformed for modelling. Niger Delta application cases were excluded from model training. The accompanying README describes the variables, preprocessing conventions, and interpretation of the dataset. The dataset can be used to reproduce the machine-learning workflow described in the associated manuscript.
Institutions
- Rivers State UniversityRivers State, Port Harcourt