In the field of sports, there are many advanced technologies to help athletes improving their skills or monitoring their body health. The force plate is one of the devices which can be used to observe the status in neuromuscular function and fatigue of athletes by (1) asking them do the countermovement jump (CMJ) on it and (2) evaluating some performance indicators such as flight time and net pulse. However, the force plate is of low portability due to it high price and heavy weight. In this paper, we propose a machine learning-based method to measure CMJ performance with an inexpensive wearable inertial measurement unit (IMU). Based on the measured acceleration, we first extract some features and then adopt the machine learning algorithm to learn several models to estimate the above indicators, respectively. The experiments are conducted by 280 countermovement jumps performed by 14 healthy subjects. The experimental results show that the proposed system is of error rate less than 8%.