Connecting tax-risk analysis to officer action
Data, rules and machine learning, connected to the people responsible for audit selection.

- Programme funder


The challenge
Income-tax and sales-tax records contain different signals of potential non-compliance. Those signals need to be reconciled, evaluated and made useful to the officers choosing cases for further scrutiny.
The assignment brought rule-based analysis and machine-learning approaches into a common decision workflow. Technical work translated tax-domain questions into risk use cases, supported validation with domain officers and addressed how the system would be used by technical teams and field formations. Training and engagement with PRAL formed part of the delivery.

Contribution and outputs
Technical work under ReMIT supported FBR’s Compliance Risk Management Wing and included PRAL training and technical handover.
- Rule-based and machine-learning risk use cases
- Technical support for audit-case prioritisation
- Domain review of identified signals
- Training and technical engagement with PRAL

