This function uses the data and the null model to generate null residuals. The null residuals are generated by first regressing random noises on the original regressors, then multiply the obtained residuals by original RSS divided by the current RSS. The results are the null residuals.
Arguments
- dat
Data frame. A data frame that used to fit the model.
- mod
Linear Model. The null model.
- test
Boolean. Whether or not to use
teston the newly generated data.
Examples
# Instantiate
x <- rand_uniform()
e <- rand_normal()
test <- vi_model(prm = list(x = x, e = e),
prm_type = list(x = "r", e = "r"),
formula = y ~ 1 + x + x^2 + e,
null_formula = y ~ x,
alt_formula = y ~ x + I(x^2))
dat <- test$gen(10)
test$null_resid(dat, test$fit(dat), test = TRUE)
#> y x e .resid .fitted test_name statistic
#> 1 3.441043 0.39981093 -0.2018527 1.1069942 2.334049 F-test 1.590544
#> 2 1.956685 0.05564954 0.6577813 -0.6577916 2.614477 F-test 1.590544
#> 3 1.147591 0.93751792 -1.4748553 -0.7483271 1.895918 F-test 1.590544
#> 4 1.187821 0.66930164 0.5465937 -0.9266438 2.114464 F-test 1.590544
#> 5 2.633396 0.23329027 1.8163460 0.1636628 2.469733 F-test 1.590544
#> 6 2.497478 0.13328545 0.5711878 -0.0537396 2.551218 F-test 1.590544
#> 7 2.470786 0.43430630 1.8795511 0.1648445 2.305942 F-test 1.590544
#> 8 1.785301 0.90614375 -1.2087911 -0.1361807 1.921482 F-test 1.590544
#> 9 4.029171 0.64952762 -0.1792063 1.8985942 2.130577 F-test 1.590544
#> 10 1.550555 0.36554707 2.3814244 -0.8114129 2.361968 F-test 1.590544
#> p_value null
#> 1 0.247652 TRUE
#> 2 0.247652 TRUE
#> 3 0.247652 TRUE
#> 4 0.247652 TRUE
#> 5 0.247652 TRUE
#> 6 0.247652 TRUE
#> 7 0.247652 TRUE
#> 8 0.247652 TRUE
#> 9 0.247652 TRUE
#> 10 0.247652 TRUE