Categorize customers to help merchants to offer promotions or discounts to cardholders.

1. Uncover signal in customer loyalty to identify the most 'valuable' individuals.
2. Create the right experience for customers. Provide specifically tailored promotions on most relevant merchant.

Data is available on Kaggle: link to Kaggle Competition.

All data is simulated and fictitious, and is not real customer data.

• train.csv - the training set
• test.csv - the test set
• historical_transactions.csv - up to 3 months' worth of historical transactions for each card_id
• merchants.csv - additional information about all merchants / merchant_ids in the dataset.
• new_merchant_transactions.csv - two months' worth of data for each card_id containing ALL purchases that card_id made at merchant_ids that were not visited in the historical data.

Number of Observations: ~ 20 million

Features/ Predicting Variables: Around 30

• Card_id - prime key
• Target_binned - loyalty category: good, bad, neutral
• Month of first purchase
• Authorized_flag - Y: Approved, N: Denied
• Purchase_amount
• State_id
• Merchant_id