Quantum algorithms represent a promising choice for solving combinatorial optimization problems, which are crucial in different real-world applications. Quadratic Unconstrained Binary Optimization (QUBO) model is the most suitable formulation for exploiting quantum solvers such as Grover Adaptive Search (GAS), Quantum Approximate Optimization Algorithm (QAOA) and Variational Quantum Eigensolver (VQE). However, quantum solvers performance can vary significantly across the different problems and, in general, cannot be determined a priori. Therefore, this thesis introduces a regression model that predicts solver performance, enabling solver selection without exhaustively running all candidate solvers on each problem instance, thus reducing computational cost and supporting hybrid quantum workflows.