TY - JOUR
AB - Given a set of data points pi = (xi, yi ) for 1 ≤ i ≤ n, the least median of squares regression line is a line y = ax + b for which the median of the squared residuals is a minimum over all choices of a and b. An algorithm is described that computes such a line in O(n 2) time and O(n) memory space, thus improving previous upper bounds on the problem. This algorithm is an application of a general method built on top of the topological sweep of line arrangements.
AU - Edelsbrunner, Herbert
AU - Souvaine, Diane
ID - 4064
IS - 409
JF - Journal of the American Statistical Association
SN - 0003-1291
TI - Computing least median of squares regression lines and guided topological sweep
VL - 85
ER -