Prescriptive Analytics in the Python Ecosystem with Gurobi [PyCon DE & PyData Berlin 2024]

Gurobi is a commercial mathematical optimization solver that guarantees globally optimal solutions for complex business problems, unlike heuristic approaches

Key use cases include supply chain logistics, airline fleet scheduling, portfolio optimization, conference scheduling, and agricultural route planning

The tool works with three main components:

• Decision variables (what you want to optimize)
• Constraints (rules/limitations)
• Objective function (what to maximize/minimize)

Free for academic and non-profit use, with commercial licenses available. Integrates with common Python data science tools like Pandas

Offers specialized packages:

• Gurobi Optimodes (pre-built optimization models)
• Gurobi Machine Learning (integrates trained regressors)
• Gurobi Pi Pandas (data-driven optimization API)

Uses absolute value (L1 norm) instead of squared error (L2 norm) for more robust optimization against outliers

Handles complex constraints like binary/integer variables, inequalities, and trained ML model integration

Differentiates from predictive analytics by focusing on optimal decision-making rather than just forecasting

Solves problems through C-based implementation with specialized mathematical algorithms developed by R&D team

Can handle large-scale optimization problems that open-source solvers struggle with, offering better performance and reliability