regDiag: Regression diagnostic

In tengku-hanis/niceFunction: These are random nice functions that I found

Description

regDiag is for screening of outliers and influential cases.

Usage

regDiag(model)

Arguments

model

lm or glm model

Value

Outlier : leverage, Studentized deleted residual (SDR)

Influential case : DFFits, DFBetas, cook's distance (cook.d)

also noted that we use dfbetas (scaled/standardized) NOT dfbeta function

TRUE : outlier and influential case present,

FALSE : no outlier and influential case

Cut-off values:

leverage > 0.2

SDR >= 3 or <= -3

DFFits > 1

DFBetas > 1

cook.d > 0.5

Examples

mod <- lm(Sepal.Length ~ Species + Sepal.Width, data = iris)
regDiag(mod)

tengku-hanis/niceFunction documentation built on Oct. 27, 2021, 7:41 a.m.