The JustNow ARS uses Machine Learning to capture and process consumer-market information that supports decision making at the retailer and provider level. While competitive ARS models are typically limited to stock data, JustNow includes sales and promotional offer data for clearer insight into the food resource management situation as well as the user behavior when making decisions to claim offers. This should result in a 10% improvement over the competition. Shopping behavior-wise, knowing the historic acquisition information of a user, their shopping decision time of the day as well as the locations in which they make those decisions, enables JustNow to send the right app triggers with the best content at the best time, thus clearing stocks faster and maximizing profit.
Architecture, challenges and solutions:
The major hurdle for this project was preprocessing and cleaning the input data in order to get a usable dataset. We had to combine a variety of data sources such as sales and stock records, promotional campaigns, weather information and national and local holidays. We had to take into consideration data on expired, damaged, stolen or lost products. Last but not least, we had to gather information on all the products including base price, shelf life, product type and tags.
