Research of a New Method for Solving Linear Regression

DOI

10.2991/tlicsc-18.2018.78How to use a DOI?

Keywords

New regression model; Least squares method; Total absolute error; Precision accuracy.

Abstract

In this paper, we propose two new models for the absolute error of a linear regression. One of which is for the average value of all data points connected the slope, and then find their average value can be obtained slope, finally use the mean point can get a linear regression equation. The other is to use the regression line after the mean value, and each point deviates from the slope of the regression line to get the average value, that is, the slope of the regression line. The least squares method is sensitive to outliers, so in the case of outliers, the new method outperforms the least square method under certain circumstances, and in some cases, the absolute error and the prediction error are better than the least square method.

Copyright

© 2018, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Yang Yu
PY  - 2018/12
DA  - 2018/12
TI  - Research of a New Method for Solving Linear Regression
BT  - Proceedings of the 2018 International Conference on Transportation & Logistics, Information & Communication, Smart City (TLICSC 2018)
PB  - Atlantis Press
SP  - 484
EP  - 488
SN  - 1951-6851
UR  - https://doi.org/10.2991/tlicsc-18.2018.78
DO  - 10.2991/tlicsc-18.2018.78
ID  - Yu2018/12
ER  -