A trade-off arises when improving one outcome means accepting a cost elsewhere. The right balance depends on the problem, the consequences of each choice, and what the solution must preserve. section 6.7.2 section 4.2.1
Trade-offs shape both computational choices and decisions under uncertainty:
- In dynamic programming, storing more subproblem results can reduce repeated computation but uses more memory. Memoization may save space when few states are reached; tabulation can use more predictable, potentially greater memory. If a recurrence needs only the previous row, earlier rows may be discarded, though recovering the decisions behind an optimal value may require extra information. section 6.7.2
- In hypothesis testing, a smaller significance level lowers the risk of a false alarm but generally raises the risk of missing a real process shift. A manufacturer weighs those errors against the costs of unnecessary interruptions and of shipping products with incorrect fill volumes. section 4.2.1
- A soft-margin support vector machine accepts some training misclassifications in exchange for a wider margin, balancing fit against model complexity. section 1.3










