# 15-feature transaction schema v2.
# Changes from v1: merchant_id scaled to 10K, device_hash to 2K.
# Within-transaction ordering grouped by semantic family (D14) so the 3-wide
# conv windows see coherent feature neighborhoods.
# Reserved token convention per feature (D6): 0=MASK, 1=OOV, 2=NULL, 3+=values.
# vocab_size = num_values + 3 for every feature.

num_features: 15
num_transactions: 64
reserved_tokens:
  MASK: 0
  OOV: 1
  NULL: 2
  values_start: 3

features:
  # --- Temporal block (indices 0-3) ---
  - name: hour
    family: temporal
    type: categorical
    num_values: 24
    vocab_size: 27
    description: Hour of day (0-23)

  - name: dow
    family: temporal
    type: categorical
    num_values: 7
    vocab_size: 10
    description: Day of week (0=Monday)

  - name: days_since_last
    family: temporal
    type: bucketed
    num_values: 30
    vocab_size: 33
    bucket_range: [0, 365]
    bucket_method: quantile
    description: Days since previous transaction, bucketed

  - name: is_recurring
    family: temporal
    type: binary
    num_values: 2
    vocab_size: 5
    description: Whether this is a recurring/subscription payment

  # --- Merchant block (indices 4-7) ---
  - name: mcc
    family: merchant
    type: categorical
    num_values: 100
    vocab_size: 103
    description: Merchant category code (100 common MCCs)

  - name: merchant_id
    family: merchant
    type: categorical
    num_values: 10000
    vocab_size: 10003
    description: Synthetic merchant identifier (10K merchants)

  - name: customer_merchant_count
    family: merchant
    type: bucketed
    num_values: 20
    vocab_size: 23
    bucket_range: [0, 500]
    bucket_method: quantile
    description: How many times this customer has transacted at this merchant

  - name: entry_mode
    family: merchant
    type: categorical
    num_values: 5
    vocab_size: 8
    values:
      0: card_present
      1: card_not_present
      2: contactless
      3: chip
      4: manual_key
    description: How the card was presented

  # --- Risk + identity block (indices 8-14) ---
  - name: amount
    family: risk_identity
    type: bucketed
    num_values: 256
    vocab_size: 259
    bucket_range: [0.01, 25000.0]
    bucket_method: quantile
    description: Transaction amount in USD, quantile-bucketed into 256 bins

  - name: card_product
    family: risk_identity
    type: categorical
    num_values: 10
    vocab_size: 13
    values:
      0: basic_debit
      1: basic_credit
      2: rewards_credit
      3: premium_credit
      4: platinum_credit
      5: business_debit
      6: business_credit
      7: prepaid
      8: virtual
      9: corporate
    description: Card product type

  - name: country
    family: risk_identity
    type: categorical
    num_values: 50
    vocab_size: 53
    description: Transaction country code (50 most common)

  - name: avs
    family: risk_identity
    type: categorical
    num_values: 5
    vocab_size: 8
    values:
      0: full_match
      1: zip_match
      2: address_match
      3: no_match
      4: not_checked
    description: Address verification result

  - name: cvv
    family: risk_identity
    type: categorical
    num_values: 3
    vocab_size: 6
    values:
      0: match
      1: no_match
      2: not_provided
    description: CVV verification result

  - name: device_hash
    family: risk_identity
    type: categorical
    num_values: 2000
    vocab_size: 2003
    description: Device fingerprint hash bucket (2000 synthetic devices)

  - name: customer_tenure
    family: risk_identity
    type: bucketed
    num_values: 10
    vocab_size: 13
    bucket_range: [0, 120]
    bucket_method: uniform
    description: Customer account age in months, bucketed