Government Data Science Teams

A Framework for Implementing Strategic Monitoring Solutions

Abstract

This paper presents a framework developed by Pernambuco’s Strategic Monitoring Data Science Team to design and implement data-driven solutions for monitoring public policies. Using an action research methodology, the study integrates data science, iterative development, and stakeholder engagement. Despite being major producers and consumers of data, governments still face significant challenges in applying data science for policy monitoring, including data quality issues, legal constraints, and institutional silos. Existing frameworks such as CRISP-DM, Scrum, and Kanban are either too technical or primarily focused on software development rather than the policy-driven decision-making required in government settings. The Strategic Monitoring Team was established within Pernambuco’s Secretariat of Strategic Projects, comprising a Chief Data Scientist as team leader, a Project Manager, three Data Scientists specializing in modeling, engineering, and visualization, and a Trainee. The team operates through an iterative five-step process: Diagnose, which involves meetings with stakeholders to identify policy issues; Plan, where internal discussions define solutions; Act, which includes the development of dashboards, reports, and applications; Evaluate, to review whether the solutions address policy needs; and learn, focusing on documenting findings and improving tools. To enhance their workflow, the team adapted Scrum methodology by incorporating policy research alongside software development, tracking projects via Notion, and deploying solutions using R and Shiny Proxy. The study highlights that traditional frameworks such as Scrum and CRISP-DM require adaptations to effectively integrate research aspects and government governance structures. By bridging data insights with decision-making processes, the team successfully balances software development, policy research, and institutional needs. The findings emphasize the necessity of specialized data science frameworks tailored for government applications, ensuring a structured yet flexible approach to strategic policy monitoring through data-driven
solutions.