By Xiaofei Wang and Bohan Huang, current students at the University of Sydney Business School, and participants in the Big Data Competition. The Competition, run by the Business School in partnership with Woolworths, challenged students to use real time data about the sales of some of their most popular product lines to predict weekly sales.
The Big Data Competition was a great opportunity to apply our business knowledge to predict product sales at different Woolworths stores. It offered us a taste of data analysis, and helped us to focus on what we really want by allowing us to apply our knowledge in real business practice.
As members of one of the winning teams (out of a total of 271 teams and 696 participants), we'd would like to share some of our reflections:
Our team members came from diverse backgrounds and areas of specialisation. We used our complementary academic knowledge to generate different ideas, as well as support different views. In our team, Xiaofei Wang was responsible for summarising our viewpoints, data testing and model application. Bohan Huang was good at finding patterns and building the models. Donghua Han focused on correlation coefficiency analysis and hypothetical testing. We let each member do their part and learned from each other. We carefully considered whether each idea was effective. Our team combined different options, gradually improved them and finally made them feasible.
We also kept a reasonable level of competition in the team. For example, we shared and discussed our views, debating over each carefully. We often first established different models, then we compared them. To choose the most appropriate model, we discussed not only the test results, but also the advantages of the models. Through these processes, we remained efficient and a competitive position.
Application of business knowledge
There were two accounting students and one commerce student in our team. However, no one in our team had a data mining and analysis background, or experience before this competition. Compared to business analysis or quantitative business students, our advantages were our understanding of business from the financial perspective, as well as our Excel skills. For example, analysing financial statements and calculating financial ratios helped us to explore the most appropriate models. Our weakness was a lack of quantitative business knowledge. So we gradually built a series of models that were simple and easy to integrate with each other, and most importantly, suitable for us to use. In the end, the models proved accurate and effective.
In the first round, we tried a variety of models, but none of them were satisfactory and we were lagging behind another team. But our team never gave up. We set a goal for ourselves and built quantitative knowledge as we went. We looked for appropriate models and improved the results. We fully used the submission opportunities and successfully achieved the top position in the first week on the public board. This moment largely encouraged us and helped us to keep trying to achieve the best performance we could, ultimately leading to our overall success.