Generation flow
Each synthetic record is built from attribute weights, selection boosts, and a revenue mapping.
1) Sample every attribute
Each customer gets a value for every attribute group. Values are chosen using the configured weights.
2) Apply selection boosts
Any value you select in setup is given a higher chance of being sampled for that attribute.
3) Add conversion bias
Higher conversion rates nudge purchase frequency and engagement toward “high.”
4) Score and normalize
Attribute weights are summed into a score and normalized to a 0–1 range.
5) Map to revenue
Revenue is mapped linearly into the min/max range with a small additive noise band.
6) Label targets (optional)
If you choose a threshold objective, each record is tagged as target when revenue meets the threshold.
