Experimental and Machine Learning Dataset for Inverse Prediction of Ternary Diesel-Biodiesel-Alcohol Fuel Composition

This dataset contains the experimental and machine learning data supporting the study “Inverse Machine Learning and Multi-Output Neural Networks for Predicting Diesel-Biodiesel Compositions in Ternary Fuel Blends via Physicochemical Properties.” The dataset includes the volumetric compositions of ternary diesel-canola biodiesel-alcohol blends containing ethanol or propanol, together with experimentally determined density and dynamic viscosity values at different temperatures. The machine learning dataset contains density, dynamic viscosity, and temperature as input variables and diesel and biodiesel volume fractions as target variables. The workbook also includes the experimental density and viscosity datasets, the validation framework used in the machine learning analysis, and a README sheet describing the variables, units, and dataset structure. The data support inverse machine learning analysis for fuel composition identification and simultaneous multi-output prediction of diesel and biodiesel fractions.