Model-Independent Reconstruction of Late-Time Cosmic Acceleration in Finsler–Randers Cosmology

• A Finsler–Randers cosmological framework is developed by incorporating anisotropic geometric corrections through a model-independent parametrization of the deceleration parameter q(z). • Cosmological parameters are constrained using a joint analysis of Cosmic Chronometers (CC), DESI BAO, and DES Y5 Type Ia supernova datasets. • The combined dataset yields the best-fit parameters H 0 = 68.16 ± 0.83 km s − 1 Mpc − 1 , η = 1.202 ± 0.017 , and ζ = 1.550 ± 0.032 . • The reconstructed expansion history gives present-day values q 0 = − 0.62 ± 0.14 and ω 0 = − 0.746 ± 0.092 ,indicating a late-time accelerated expansion. • Higher-order cosmological diagnostics and information criteria indicate that the Finsler–Randers model provides a viable geometric alternative for describing late-time cosmic acceleration. We investigate a Finsler–Randers cosmological model that introduces anisotropic geometric corrections to the expansion dynamics through a model–independent parametrization of the deceleration parameter q ( z ). The cosmic expansion history is reconstructed using observational data from Cosmic Chronometers (CC), DESI baryon acoustic oscillations (BAO), and DES Y5 Type Ia supernovae. The combined analysis constrains the model parameters to H 0 = 68.16 ± 0.83 km s − 1 Mpc − 1 , η = 1.202 ± 0.017 , and ζ = 1.550 ± 0.032 . The reconstructed expansion history yields present-day values q 0 = − 0.62 ± 0.14 and ω 0 = − 0.746 ± 0.092 , confirming a late-time accelerated phase consistent with observations. Higher-order cosmological diagnostics, including the jerk and snap parameters, together with the statefinder and Om diagnostics, indicate a dynamical dark energy behavior arising from geometric anisotropic corrections. Information criterion analysis shows that the model provides a statistically viable description of the late-time expansion when compared with the ΛCDM scenario. These results suggest that anisotropic geometric effects within the Finsler–Randers framework can reproduce the observed cosmic acceleration without introducing a cosmological constant.