<p>Latent heat thermal energy storage systems offer high energy storage density at near uniform temperature. However, they suffer from the challenge posed by low thermal conductivity of the solid-liquid phase change materials. Electroconvection induced by the application of electric field is a promising heat transfer enhancement technique to overcome this limitation. Here, we present a data-driven approach to study electrohydrodynamic-assisted melting. The electric Rayleigh number (T), the Rayleigh number (Ra), and the Stefan number (St) are fed as inputs to the data-driven machinery. First, numerical simulations are performed considering a computational domain of a square differentially heated cavity with a circular electrode. Data from numerical simulations is then used to train an artificial neural network. The artificial neural network captures the relationship between the nondimensional numbers and the growth of the total liquid fraction. Finally, the trained neural network is used to generate a large number of samples for a global sensitivity analysis at a low computational cost. Sensitivity indices computed across time provide insights on the role played by electroconvection, buoyancy-driven convection, and conduction on the melting rates. For the input parameter space considered herein, the sensitivity indices of the melting rates to electric Rayleigh number and Rayleigh number remain negligible till the total liquid fraction reaches 0.2. The trend of the first and total-order sensitivity indices for electric Rayleigh number reflects the impediment caused by the presence of the electrode on electroconvection. Second-order sensitivity indices are non-trivial only for the pair of electric Rayleigh number and Rayleigh number suggesting coupling between electroconvection and buoyancy-driven convection.</p>