Explain influence analysis and the techniques used to spot observations that sway a regression unduly

Some observations pull the fitted line much harder than others. Leverage measures how far an observation's independent variables sit from their means, between 0 and 1; flag values above 3 × (k + 1) ÷ n. An outlier is extreme in the dependent variable: its studentized residual, computed from a model estimated without it, exceeds the critical t-value. An influential point is one whose removal would change the estimates materially, and Cook's distance combines leverage and residual size to measure that, with values above 0.5 worth a look, above 1 likely influential, or above twice the square root of k ÷ n. High leverage alone is not influence: a point far out but on the line changes little.