Regression analysis is a useful tool in many fields, with the least squares ratio method being better than the ordinary least squares method, especially with outliers. A new method called the weighted least squares ratio was introduced to improve M-estimators. The study compared this new method with the weighted least squares method to see which one performs better when dealing with outliers and variance in regression models. The results showed that the weighted least squares ratio method had lower mean absolute errors in estimating regression parameters and dependent values.