Residualkriterier

Residualkriterier refers to a set of criteria and diagnostic rules used to evaluate the residuals of a statistical model, primarily in regression and time-series analysis. Residuals are the differences between observed values and model-fitted values, and residualkriterier help determine whether a model adequately captures the underlying relationships.

The purpose of residualkriterier is to assess whether model assumptions hold, identify departures such as nonlinearity, heteroskedasticity, and autocorrelation, and detect outliers or influential observations. They guide decisions about model revision, transformations, or the use of robust methods.

Common components of residualkriterier include visual and numerical tools. Visual diagnostics involve plots of residuals versus fitted values to inspect patternlessness, and QQ plots to judge normality of residuals. Statistical tests target specific assumptions: tests for normality (for example, Shapiro–Wilk), tests for homoskedasticity (such as Breusch–Pagan or White test), and tests for autocorrelation (like Durbin–Watson or Ljung–Box). Influence and outlier detection use measures such as standardized or studentized residuals, Cook’s distance, and leverage statistics. Fit quality can be summarized with metrics like RMSE, MAE, and, in context, R-squared.

Residualkriterier are applied during model diagnostics, validation, and refinement. They emphasize that passing criteria do not guarantee causal validity and that results depend on model specification, sample size, and chosen methods. They are integral to ensuring transparent assessment of model reliability and robustness.