Machine learning Up until recently , any sales forecasting was done by humans by looking at previous data but it was an imperfect science as humans are not only swayed by previous experience , external biases and other factors , but they can ’ t handle large volumes of data . Alexander Khaytin , COO at machine learning and data analytics company Yandex Data Factory says : “ Machine learning can help retailers to deliver accurate forecasts for product sales . By applying predictive and recommendation models to historical sales data , retailers are able to forecast at a granular level – by product , individual store and even day of the week . This helps to optimise supply chain management , lowering risks of overestimation or underestimation for product inventory and increase efficiencies across the whole business . “ For example , Russian retailer Pyatyorochka piloted a specialised machine learning model that forecasted demand for individual products on sales promotions from its different stores . The project resulted in Pyatyorochka forecasting the exact number of wholesale packages needed 61 percent of the time , and was correct within one package 87 percent of the time . A phenomenal percentage when compared to those produced through human analysis .”

SEVEN EMERGING FUTURE TECHNOLOGIES

IS MACHINE LEARNING A GAME CHANGER IN MARKETING ? Perspectives and limitations . A panel discussion moderated by Norbert Wirth ( GfK , Global Head of Data and Science ) Participants : Andreas Braun ( Allianz , Head of Global Data and Analytics ) – Martin Szugat ( Predictive Analytics World Germany , Program Chair ) – Raoul Kübler ( Ozyegin University , Istanbul ).

25