morie_survnnr_fit

A step of the survnnr_native implementation. No other function in the package calls it. See the file header for the source the module follows. the source it follows.

Usage

morie_survnnr_fit(
  X,
  times,
  events,
  hidden = c(),
  activation = "tanh",
  l2 = 0,
  lr = 0.1,
  n_epochs = 400,
  seed = 0,
  tol = 1e-10
)

Arguments

X

A vector; its length is taken and its elements indexed.

times

A vector; its length is taken.

events

A vector; its length is taken.

hidden

A vector; its length is taken. Defaults to c().

activation

Passed to morie_survnnr_forward. Defaults to "tanh".

l2

Numeric; combined arithmetically in the body. Defaults to 0.

lr

Numeric; combined arithmetically in the body. Defaults to 0.1.

n_epochs

Coerced to integer by the body, with as.integer. Defaults to 400.

seed

Passed to .survnnr_init. Defaults to 0.

tol

Passed to <. Defaults to 1e-10.

Value

A list with estimate, W, b, activation, hidden, l2, loss_history, risk, centred_risk, coefficients, times, events, epochs, ties, scale_note, method.

Examples

set.seed(1); n <- 40
X <- lapply(seq_len(n), function(i) rnorm(3)); tm <- rexp(n); ev <- rbinom(n, 1, .7)
fit <- morie_survnnr_fit(X, tm, ev, hidden = c(4L), n_epochs = 20)
fit$risk
#>  [1] -0.32640504 -0.60752765 -0.31352296 -0.78335959  1.06495740  0.32448003
#>  [7] -0.06223376 -0.50703415 -0.09953326  0.54851534  0.10621854  0.73067078
#> [13]  0.41448858 -0.07688100 -0.63732879  0.11705313 -0.45815199 -0.42508112
#> [19] -1.05628547 -0.30963818  0.17095295  0.54771506 -0.54025078 -0.61397448
#> [25]  0.07432222  0.27620593  0.17481015 -0.66410312  0.13833775 -0.13422592
#> [31] -0.31938237 -0.77603448 -0.08786969  0.48366342 -0.25361076 -0.11744949
#> [37] -0.96878400 -0.75150303  0.16627580 -0.37677602
fit$epochs >= 2
#> [1] TRUE