A resilient data-driven predictive control algorithm without terminal components is proposed for an unknown linear time-invariant system under DoS attacks. Compared with the traditional model predictive control, this paper learns the behavior pattern of the system by analyzing the input and output data of the system, and predicts the future input and output, only relying on the historical input and output data of the system. When DoS attacks occur, the scheme can use the characteristics of predictive control to compensate for the lost data caused by the DoS attack, thereby reducing the negative impact of the DoS attack on the system. Then we prove that the proposed scheme can guarantee the robust stability of the system considering bounded network induced noise and process noise. Finally, the effectiveness and feasibility of the scheme are verified by numerical simulation results. The experimental results show that the scheme has stronger robustness and anti-DoS attack ability.