dunnhumby & hack/reduce: A Product Launch Challenge!
Saturday, May 11, 2013 from 9:00 AM to 9:00 PM (EDT)
The importance of early feedback on new product launches
Grocery stores are always looking to introduce new products, such as a new flavor of drink, a new type of yogurt, or the latest DVD. In the highly competitive world of retailing, the success or failure of these new product launches is highly dependent on the first few weeks of sales. The earlier you know how well the product launch is doing, the earlier you can decide whether you want to increase advertising and the number of stores selling the product, or whether you cut your losses on a product that is doomed to fail.
The mission: Predict how successful each of a number of product launches will be 26 weeks after the launch, based only on information up to the 13th week after the launch.
The data: In order to create and train your forecasting models, we will provide a training set which includes the information for the full 26 weeks, for around 3000 product launches. For the challenge, the data set will contain 1000 product launches over 13 weeks, and we ask that you predict the number of units sold for each of these launches in week 26. We have reduced the data to a modelling set of tens of thousands of rows in order for the participants to focus on the predictive modelling rather than preparing the data.
The training set and question set each contain by week for each launch:
- The category of the product, such as Bread, Coffee or Video Games
- The number of stores selling the product (note that this for all 26 weeks in the question set rather than just the first 13 weeks)
- The number of units sold that week
- The number of distinct customers who have bought the product (cumulative)
- The number of distinct customers who have bought the product at least twice (cumulative)
- Cumulative units sold to a number of different customer groups: Convenience at home, Family Focused, Finest, Grab and Go, Shoppers On A Budget, Traditional Homes, Watching the Waistline, Least Price Sensitive, Price Sensitive, Splurge and Save, and Very Price Sensitive
Who should attend?
Individuals with a passion for data science, data modeling, and statistics. Participants are welcome to form teams or work individually for this competition.
How will I know how well I'm doing?
We will be using the Kaggle platform for this competition, making it easy to know your ranking throughout the day.
What if I have a question?
Domain and data experts will be on hand to answer any questions that you have, just let us know!
And, of course, who wins?
We will score each of your predictions based on how close you are to the actual result. An exact prediction scores you the maximum 100 points, and the further you are away from the correct prediction, the less points you score. We sum your scores across all the launches, and whoever scores the most wins!
We have $4,000 in cash prizes and $1,000 in gift cards to distribute across the first, second, and third place winners!
dunnhumby is the world’s leading customer science company. We analyze data and apply insights from more than 400 million customers across the globe to create better customer experiences and build loyalty. Our insights and strategic process help clients create competitive advantage and enjoy sustained growth.
hack/reduce is a non-profit established in partnership with the State of Massachusetts, a number of local and global firms committed to innovation, and in collaboration with MIT, Harvard and other local universities. Working closely with our partners and the community, we’ll bring developers, data scientists and domain experts across disciplines together to create the next generation of Big Data technologies and applications.
Please direct any questions you may have to email@example.com.
Getting to hack/reduce: 5 minute walk from the Kendall T stop. Just take 3rd street down 3 blocks to the building on the left in the Kendall Boiler & Tank Building. Street parking around the building and numerous garages including 350 Kendall Street.
When & Where
hack/reduce // Boston's Big Data hackerspace