Sie sind auf Seite 1von 1

DIAGNOSTIS FOR REGRESSION

--HETEROSCEDASTICITY TEST

mytest=lm(y~x+x1, data=mydata) first run the regression with a name,


bptest(mytest) this code enough to apply Breusch-Pagan to the model,
bptest(mytest~x+ I(X)^2, data=mydata) runs the white test for the model.
TESTING FOR OMITTED VARIABLE
RAMSEYS RESET TEST
Resettest(mytest) this should yield if the model has any misspesification
TESTIN FOR AUTOCORRELATION and TS REGRESSION
library("dynlm")
mytest=dynlm(y~x+L(x)+d(x), data=mydata),
dwtest(mytest) Durbin-Watson test

bgtest(mytest) Breusch-Godfrey test for serial correlation of order 1


vcov(mytest) or vcovHC(mytest) this produces HAC residuals
coeftest(mytest, vcov=vcovHC) this gives coeeficients of the model regarding
Whites HAC residuals.
Testing for stationarity in data
library(tseries)
Adf.test(diff(log(mydata))) augmented Dickey Fuller testwith Ho= data has
unit rooot.
pp.test(diff(log(mydata))) Phillips Peron test with same hypothesis
kpss.test(diff(log(mydata))) kppss test with the same hypothesis

Das könnte Ihnen auch gefallen