1-hold out logistic regression predections. Zero indexed.

xlogistic_fits(x, y, w, i, j)

Arguments

x

NumericVector, expanatory variable.

y

NumericVector, 0/1 values to fit.

w

NumericVector, weights (required, positive).

i

integer, first index (inclusive).

j

integer, last index (inclusive).

Value

vector of predictions for interval.

Examples


set.seed(5)
d <- data.frame(x = rnorm(10))
d$y <- d$x + rnorm(nrow(d))>0
weights <- runif(nrow(d))
m <- glm(y~x, data = d, family = binomial, weights = weights)
#> Warning: non-integer #successes in a binomial glm!
d$pred1 <- predict(m, newdata = d, type = "link")
d$pred2 <- xlogistic_fits(d$x, d$y, weights, 0, nrow(d)-1)
d <- d[order(d$x), , drop = FALSE]
print(d)
#>              x     y      pred1          pred2
#> 3  -1.25549186 FALSE -2.4292992  -2.167502e+00
#> 1  -0.84085548  TRUE -1.8751975 -1.511600e+308
#> 8  -0.63537131 FALSE -1.6005976  -1.581696e+00
#> 6  -0.60290798 FALSE -1.5572150  -1.394288e+00
#> 7  -0.47216639 FALSE -1.3824977  -1.150064e+00
#> 9  -0.28577363 FALSE -1.1334107  -8.816750e-01
#> 4   0.07014277 FALSE -0.6577798  -4.424278e-01
#> 10  0.13810822 FALSE -0.5669537  -2.757255e-01
#> 2   1.38435934  TRUE  1.0984811   8.595039e-01
#> 5   1.71144087  TRUE  1.5355784  -7.673982e-01