A decision variable represents a choice in an optimization problem. In linear programming, it records which alternatives are selected and appears in both the objective function and the constraints. Its coefficients connect each choice to value, cost, effort, resource limits, or strategic requirements. section 8.2
Define decision variables by specifying their meaning, allowable values, and relationships to the optimization model:
- Use binary variables when each alternative is a yes-or-no choice: 0 means excluded and 1 means selected. section 8.2.5
- Use continuous variables when a decision represents a quantity that can vary within a range. section 8.3
- Combine variables with an objective function and linear constraints so the solver can find a feasible, high-value or low-cost solution. section 8.2
Selecting project features
For a feature-selection model, define one binary variable per feature. In Python, `x_bounds = [(0, 1) for _ in features]` gives every variable the allowable range 0–1. Passing these bounds to `scipy.optimize.linprog` restricts the solution to selecting or excluding each feature while budget and epic constraints are enforced. section 8.2.5 section 8.2.2
