Uncertainty analysis enhances the precision, reliability and applicability of mathematical models. It tests the model under various scenarios to ensure validity. This research uses a machine learning algorithm to analyse uncertainty in a boundary layer model for an unsteadily stretching cylinder with variable fluid properties. The research accounts for temperature-dependent changes in viscosity and thermal conductivity and examines two distinct viscosity models. The first model is based on an inversely linear viscosity–temperature correlation. In the second model, an exponentially varying viscosity function derived from an empirical result is employed. The governing equations, reduced via boundary layer approximations, yield self-similar solutions featuring a parameter \(S\) that represents the cylinder’s stretch rate. Numerical simulations have been conducted for \(S\) values ranging from −4 to −1, considering a Prandtl number reaching up to 7. The cylinder’s stretch rate influences the flow behaviour in the following manner. For a rapidly decreasing stretch rate, the boundary layer becomes thinner, requiring a greater driving force at the boundary. In contrast to the no-slip condition, the fluid velocity at the wall is less than the wall velocity, which results in a smaller momentum diffusion. As a result, increasing the wall slip coefficient reduces fluid motion, thereby suppressing the boundary layer. This study reveals a notable difference between the outcomes for constant and variable physical properties. It is concluded that ignoring viscosity dependence on temperature leads to inaccurate numerical estimations of the flow model, when sufficiently large temperature differences are present. Subtle quantities like resisting wall shear and Nusselt number are evaluated and explored under various controlling parameters. The application of the LASSO algorithm determined that the dominant factor influencing the driving force required by the cylinder is the parameter linked to the growth rate of the stretch rate \(S\) . Additionally, the Prandtl number \(Pr\) was identified as the primary factor impacting the cylinder’s cooling rate.
