Dataset and Code Availability for Article with title "Machine Learning Prediction of Reverted Austenite in 18% Ni Maraging Steel"

The dataset aus_rev.csv contains 204 entries with eight numerical variables that describe the chemical composition, heat-treatment parameters, and resulting phase fraction in an alloy system, likely related to maraging or austenitic steel reversion studies. The columns include the elemental compositions of Nickel (Ni %), Molybdenum (Mo %), Cobalt (Co %), Iron (Fe %), and Titanium (Ti %) in weight percent, along with the heat treatment duration (Time (hrs)) and temperature (Temperature (°C)). The output variable, %fraction, represents the fraction (in percent) of a specific phase, probably reverted austenite or a precipitate phase formed during treatment. All variables are numeric and free from missing values, making the dataset well-suited for data-driven modeling or regression analysis. Overall, it captures the relationship between alloy composition and thermal processing conditions, and how these factors influence the resulting phase fraction in the material. The provided Python script builds a machine learning model to predict the volume fraction of reverted austenite in alloys based on their chemical composition and heat treatment conditions. It loads and cleans training and test datasets, applies a nonlinear feature transformation for scaling, and trains multiple regression models including Linear Regression, Decision Tree, Random Forest, and Gradient Boosting. A Voting Regressor combines these models to improve prediction accuracy. The code evaluates model performance using cross-validation, Mean Squared Error (MSE), and R² score, and visualizes the results through actual vs. predicted plots, providing a complete workflow for data-driven materials analysis.