Part of Tianchi Datamining Competition
Sponsored by Cainiao
At Alibaba, our e-commerce platform accumulated massive amount of data from both buyers and sellers. A crucial part of our e-commerce business is to help sellers to better plan their supply chain with those data. With advanced data-mining technologies, we are able to automate most part of inventory planning and achieve a highly accurate forecast, leading to lower operational cost and more efficient inventory management for sellers. We believe this technology could benefit the entire e-commerce industry. Therefore, we present this problem to the public, in search of innovative techniques to solve this challenging yet rewarding problem.
High quality demand forecast model is the core of supply chain management. In this problem, we provide e-commerce activity data for a given set of products on Alibaba’s e-commerce platform for a whole year. The challenge is required to predict the demand of those products for the next two weeks, both on a national and regional level. Contestants are asked to utilize advanced data-mining and machine learning techniques to accurately model the pattern of the demand over time and to consider the uncertainty in future demand, in order to find an optimal solution on inventory planning. A better forecast means lower operational cost, better customer experience, and therefore enhances the total efficiency of supply chain operation.
During this contest, every team should predict the optimal national and regional inventory of the next two weeks(2015.12.28-2016.01.10). Meanwhile, we will provide a config file that contains the cost(A) when the predicted inventory is less than the real sales and the cost(B) when the predicted inventory is more than the real sales. According to this file, the total cost will be calculated by the formula below. The target of each team is to achieve the lowest cost.
Fan Wu - from IECAS
Bohui Rong