What type of study is this?

This is a Quantitative Study study.

October 17, 2025

Code and Data Repository for Outlier detection in regression: Conic quadratic formulations

Key Points

The software effectively identifies outliers in regression scenarios, leveraging conic quadratic formulations.
Using Gurobi and Mosek, the implemented methods relate closely to the Least Trimmed Squares approach in statistics.
Developed in Java, the solution requires licenses for commercial solvers to run efficiently.
Outlier detection methods can improve the accuracy of regression models, making them more reliable for analysis.

Abstract

The goal of this software is to demonstrate the use of mixed-integer conic quadratic formulations to detect outliers in regression problems (closely related to the Least Trimmed Squares problem in statistics). The methods are implemented in Java and rely on commercial solvers Gurobi and Mosek. Executing the code requires a license for these solvers.

Demander à l'IA

Bookmark