Abstract: Methods, systems, and apparatus, including computer program products, for providing recommendations in a retail environment. A recommendation system includes a centralized knowledge base including one or more rules for defining logic to provide recommendations based on one or more detected system alerts, an alert engine for determining alert conditions in a retail presence including out of stock conditions, a root cause engine for determining one or more root causes for out of stock conditions that are detected, a mapping engine for mapping between alerts and recommendations, and a recommendation engine for issuing one or more recommendations associated with given alerts and determined root causes.

Abstract: Methods and systems for predicting out-of-stock occurrences and for assigning root causes to out-of-stock occurrences are described. In one implementation, inventory data and point of sale data are collected. An expected lost sales value is determined. A true demand is determined based on the point-of-sale data and the expected lost sales value. A probability of an out-of-stock occurrence is determined based on the inventory data. In another implementation, an out-of-stock occurrence is identified. The out-of-stock occurrence is classified, and one or more root causes are assigned to the out-of-stock occurrence.