DFFITS is a diagnostic measure in regression analysis used to identify influential data points that may unduly affect the fitted model. It quantifies the difference in the predicted value when a particular observation is excluded, helping analysts ensure the robustness of their model's predictions.