Strategic safety stock placement (SSP) is a tactical model to deter-mine optimal place and level of safety stocks in a supply chain. In SSP research, the guaranteed-service (GS) model approach can determine the best choice of the inventory placement to minimize total supply-chain inventory cost, while assuring 100% service-commitment under the bounded demand assumption. Given this simplicity and exactness, this modeling has become the model of choice in SSP research. Existing literature assumed that the demand bound is known or can be calculated out of the model by the confidence region of the demand prediction, which is hardly justified in applications. In this research, we propose a data-driven optimization approach for the SSP in which decision maker has no prior information about demand bound and must learn from the data. We applied predictive prescription approach in which we have the data consisting, not only of observations of demand itself, but observations of associated auxiliary quantities such as weather forecast. The problem is a conditional stochastic optimization problem in which a decision is made on the basis of an observation of auxiliary data. Numerical results demonstrate that the learning algorithm performs well on the test data motivated from industrial applications.