# Studentized residual test

Studentized residual is computed as regression model residual divided by its adjusted standard error.

Residuals are obtained by subtracting the target value that is predicted by the regression model, from observed target value for each data row. Standard error is given by the square root of the mean square for the error source. The adjustment of the standard error consists in multiplying it by the square root of leverage value that is subtracted from one. Leverage value is computed based on the design matrix and design matrix row for the data row. It adjusts the standard deviation by taking into account predictor values.

An outlier test for studentized residuals is conducted by comparing the absolute value of studentized residual with threshold value 3. Studentized residuals are distributed according to t distribution and the probability of being greater than the threshold is less than 1%.

Points with highest ranking studentized residuals above the threshold value are reported as meaningful differences, that is outliers in this case.