Robust Regression (RANSAC)

Robust Regression, specifically RANSAC (Random Sample Consensus), is an iterative model fitting algorithm used to estimate parameters of a mathematical model from a set of observed data, particularly in the presence of Outliers.

The algorithm works by iteratively selecting a subset of the data points, known as the inliers, that are most likely to belong to the underlying model, while also identifying outliers. These inliers are then used to estimate the model parameters, and the quality of this model is assessed by counting the number of inliers that fit well with the model within a certain tolerance.


  • RANSAC provides a robust estimation of model parameters, especially in the presence of a significant percentage of outliers or noise in the data.
  • RANSAC is commonly used for fitting models in situations where traditional least squares methods may be heavily influenced by outliers, such as in Computer Vision, image processing, and geospatial analysis.
  • The algorithm's performance is influenced by several parameters, including the number of iterations, the threshold for defining inliers, and the minimum number of inliers required to accept a model.
  • RANSAC is particularly suited to situations where the dataset contains a substantial amount of noisy or corrupted data points, allowing it to produce reliable model estimates even in the presence of such outliers.
  • While RANSAC is effective in mitigating the impact of outliers, it may struggle with datasets where the number of outliers is extremely high or when the inlier ratio is very low, as this can hamper the accurate estimation of model parameters.