Holiday Sales Forecasting Contest

Forecasting time series data can be part art and science. With the holiday season around the corner, all the B2C businesses are gearing up for the best sales season of the year.

Can a complex activity such as predictive analytics be crowd sourced? Crowd sourcing is a highly cost effective solution that has been very successfully deployed in various services

A new forecasting contest will challenge solvers to forecast sales for the entire holiday season at a monthly level: November 2012, December 2012 and January 2013.

Data challenge
Before you equip yourselves and jump on to the challenge, note that there is a data challenge thrown in unlike other contests. No description of the predicted variables or the predictor’s are available per client requirements.

The best forecaster wins $650. The contest will satisfy the learning buffs too. The lead analyst will share a previously successful model with the community after the contest closes.

Lowest error
Prizes will be conditionally awarded to the top 13 participants whose eligible entry produces the lowest mean absolute percentage error (MAPE) calculated against the lead analyst’s predictions.

Submission Deadline is friday 14 December 2012, the results will be announced by saturday 29 December 2012.

Read more…