This paper proposes data-driven H∞ feedback control which is synthesized based on LMI formulation. With I/O data, a closed-loop output predictor is parameterized by stochastically uncertain Markov parameters. Those Markov parameters are estimated by least squares. The estimation error due to bias and noise is minimized using H∞ approach. The state is composed of I/O data, which makes stability analysis possible. The stochastic bounded real lemma plays the key role to derive mean square stable condition and the problem is solved using numerically efficient LMI method. This research is applied to wind turbine benchmark model to demonstrate its effectiveness. The output power trajectory is successfully tracked by data-driven H∞ feedback controller.