Family
Thompson Sampling
in-developmentApache 2.0rlretail-facing
Prediction Engine.
Thompson Sampling RL agent for next-best-action prediction over user-behavior signals. Continuous online learning — no one-shot training run.
Foundation · Thompson Sampling
What it does
The honest description.
Thompson Sampling RL agent for next-best-action prediction over user-behavior signals. Continuous online learning — no one-shot training run.
Architecture
How it works.
User behavioral signals: transaction history, spending patterns, goal state, account activity. Thompson Sampling context arms (action history, reward signal). Next-best action prediction (savings goal, bill payment due date, budget alert). Continuous reward signal from user acceptance/rejection drives online policy update.
Training data & drift
What it learned from, and how we watch it.
Source
Inputs: User behavioral signals: transaction history, spending patterns, goal state, account activity. Thompson Sampling context arms (action history, reward signal).
Drift baseline · current quarter
Drift baseline not yet documented for this model.
What it's NOT
- Drift monitored via RLStrategy (reward-mean drop + variance), same as budget_guardian/savings_optimizer/bill_payment.
Version history · immutable once published
Every release, traceable.
Data Mukodi holds about you is subject to DPDP §12 erasure (use /consent in the app), and any concern about this model can be raised via [email protected] under DPDP §13 grievance, or via /security.