Optimization and Sensitivity Analysis of Least Absolute Estimator

Least absolute estimator is one of the most commonly used regression method, people concerns it because that its regression result is slightly impacted by singular points, but the so-set models are non-smoothly, which increases the difficulty of calculation. In this paper, for various observations and needs, we first transform least estimation model into different linear programming models, and then solve them by using the relevant software. At last, we set sensitivity analysis on different situations to identify the factors affecting the results further.