This is a first blog post of this marketing analytics series on a coupon e-commerce website (ponpare.jp).

Based on a year of transactional data published by a coupon e-commerce website (ponpare.jp) for this Kaggle contest, I plan to share some of the interesting marketing insights and trends in this series of blog posts. This blog posting covers a basic description of data background. In subsequent postings, I will cover topics such as product category specific patterns, customer classification through dimensionality reduction, retrospective A/B testing based on customer visits and purchases.

Brief Introduction of Data:

There are 5 main files which contain each user profile info, visit logs, transaction data and each coupon info. Some of these are divided between a training set and a test set for validation purposes as the goal of the Kaggle contest is making predictions. Depending on the type of analysis, I will clarify if I will be using one or both.

• Total # of users (User List): 22,873
• Training set: 2011-07-01 to 2012-06-23
• Test set: 2012-06-24 to 2012-06-30

Structure of Files:

The variables represented below are simplified to show the basic characteristic of the dataset and relationships between each file.

1. User List (by each User_ID)
   • User_ID
   • Sex
   • Gender
   • etc
2. Coupon Visit (visit log, shows each visit by a user for any coupon)
   • User_ID
   • Coupon_ID
   • Purchased? (1 or 0)
   • etc
3. Coupon Detail (by each transaction)
   • User_ID
   • Coupon_ID
   • Purchase_ID
   • etc
4. Coupon List
   • Coupon_ID
   • Category name
   • etc
5. Coupon Area (area in Japan where the coupon is listed)

Index of this Series:

• Intro (this post)
• Uncovering User Data
• Purchasing patterns