Abstract Symbolic regression searches for mathematical expressions that best fit a given set of data points. In this work we use this regression method to describe phase transformations from austenite to product phases for different isothermal and non-isothermal cooling conditions with a single analytical equation. Besides the data processing steps, we describe the underlying mathematical algorithm and present an expression for the employed data set. The resulting equation describes the phase transformation rate and can be used to describe the phase transformation kinetics for arbitrary cooling paths. The result is validated on four additional measurements with different heat treatment and final microstructures. The application of the method to a wider range of chemical compositions is discussed.
Bloder et al. (2026) studied this question.