Quantum reservoir computing is a type of machine learning in which the high-dimensional Hilbert space of quantum systems contributes to performance. In this study, we employ the Bose-Einstein condensate of dilute atomic gas as a reservoir to examine the effect of reduction in the number of condensed particles, damping, and the nonlinearity of the dynamics. It is observed that for the condensate to function as a reservoir, the physical system requires damping. The nonlinearity of the dynamics improves the performance of the reservoir, while the reduction in the number of condensed particles degrades the performance.