Skip to contents

Sample Nonparametric Ordinal Curve Data

Source: R/nonparametric-ordinal-plot.R
sample_nonparametric_ordinal_data.Rd

Simulates pre-computed grade-specific probability curves matching what SAS produces after fitting a cumulative proportional-odds nonparametric temporal trend model. Covers the ordinal TR / AR grade patterns from:

  • tp.np.tr.ivecho.average_curv.ordinal.sas (p0, p1, p2, p3 individual probs)

  • tp.np.po_ar.u_multi.ordinal.sas (multi-scenario ordinal, p34 = p3+p4)

Usage

sample_nonparametric_ordinal_data(
  n = 1000,
  time_max = 5,
  n_points = 500,
  grade_labels = c("Grade 0", "Grade 1", "Grade 2", "Grade 3"),
  n_bins = 10,
  seed = 42L
)

Arguments

n

Number of simulated patients (controls CI width and data-point variability). Default 1000.

time_max

Upper end of the time axis (years). Default 5.

n_points

Number of time points on the prediction grid. Default 500.

grade_labels

Character vector of grade labels, one per grade level in ascending order. Default c("Grade 0", "Grade 1", "Grade 2", "Grade 3").

n_bins

Number of equal-sized bins for the data summary points. Default 10.

seed

Random seed. Default 42.

Value

A long-format data frame: time, estimate, grade (factor). Individual grade probabilities sum to 1 at each time point.

Details

SAS context: The SAS predict dataset has one column per grade (p0, p1, p2, p3) after computing individual probabilities from cumulative probabilities (cp0 = co0/(1+co0), etc.). Export it to CSV and reshape to long format to use your own model output.

See also

hv_ordinal(), sample_nonparametric_curve_data(), sample_nonparametric_ordinal_points()

Examples

dat <- sample_nonparametric_ordinal_data(n = 800, time_max = 5)
head(dat)
#>         time  estimate   grade
#> 1 0.01000000 0.6276258 Grade 0
#> 2 0.01012532 0.6276898 Grade 0
#> 3 0.01025221 0.6277545 Grade 0
#> 4 0.01038069 0.6278200 Grade 0
#> 5 0.01051078 0.6278864 Grade 0
#> 6 0.01064250 0.6279535 Grade 0
# verify probabilities sum to 1 at each time point
tapply(dat$estimate[dat$time == dat$time[1]],
       dat$grade[dat$time == dat$time[1]], sum)
#>    Grade 0    Grade 1    Grade 2    Grade 3 
#> 0.62762585 0.22076622 0.10053344 0.05107449