Product availability is an important component to maintain consumer satisfaction and secure revenue streams for the retailer and the product supplier. Empirical research suggests that products missing from the shelf, also called 'out-of-shelf', is a frequent phenomenon. One of the challenges is to identify products missing from the shelf on a daily base without conducting physical store audit. Through empirical evaluation, this study compares various classification algorithms that can identify 'out-of-shelf' products, which is the minority class of product availability. Due to the class imbalance of product availability, an ensemble learning method is used to increase performance of the base classifiers used. The validation results indicate that it is possible to deliver accurate predictions regarding which products are 'out-of-shelf' for a selected retail store on a daily base. However, the predictions could not identify a significant number of the products missing from the shelf.