Practical Optimization for Stats Nerds
summary
Shows how familiar estimators are reformulations for solvers and then actually builds them: least squares via closed form, via numpy, and again as a quadratic program using CVXOPT QP; SVM as a linear program solved with pulp mixed‑integer variables; k‑means with binary assignment variables. Demonstrates a portfolio example and a ride‑pooling (pedicab) model with two riders per vehicle, generates 2387 feasible routes, reports an objective cost of 814, and uses column generation and a 10% margin constraint to shorten routes.
AI-written summary · about
chapters (8)
- 0:00Optimization for data scientists
- 4:14Least squares regression
- 15:07General optimization models
- 16:57Portfolio optimization
- 20:57SVMs via linear optimization
- 26:45Clustering with integer programming
- 31:00Ride-sharing route optimization
- 35:37Q&A
related talks
01h 28mFollow the gradient: an introduction to mathematical optimisation
19mML ain’t your only hammer: adding mathematical optimisation to the data scientist’s toolbox
01h 15mEducational framework for Black Box optimization methods design
01h 10mAnalysis and Design of Optimization Algorithms via Integral Quadratic Constraints
18mProduct Management for Optimization
01h 03mSampling for Linear Algebra, Statistics, and Optimization I
01h 18mOptimization I
43mComputational and Statistical Tradeoffs via Convex Relaxation