library(stats) # needed for the stepfun method

# For using the boot function:
#
samplemedian = function(x, d){
    return(median(x[d]))
}

samplemean = function(x, d){
    return(mean(x[d]))
}

make_kaplan_meir_survival_curve = function(DF){
    #
    # Build/compute a S(t) curve from the formual given in 11.4.1
    #
    # In calling this routine we assume that we have cosumn dataframe named:
    # 1) deceased: 0 => deceased and 1 => alive
    # 2) time: lifetime
    #
    #-----

    # Drop any d_i == 0 points (these don't affect the estimate of S(t)):
    #
    DF = DF[DF$deceased == 1, ]

    n = dim(DF)[1]

    # Order the dataset by time:
    #
    inds = order(DF$time)
    DF = DF[inds,]

    # Drop duplicate records (by time):
    #
    duped = duplicated( DF$time )
    DF = DF[!duped, ]

    possible_t_values = DF$time
    n_ts = length(possible_t_values)

    # Compute the midpoints between each time:
    mids = ( possible_t_values[-1] + possible_t_values[-n_ts] )/2
    is = 1:length(mids)
    cps = cumprod( ( (n_ts - is) / (n_ts - is + 1) ) )

    xs = DF$time # the jump locations
    ys = c( 1, cps, 0 )

    library(stats)
    sf = stepfun(xs, ys, right=TRUE)

    return(sf)
}

load_table_11_5_data = function(){
    # The Treatment Time data: Table 11.5
    # 
    DF = data.frame( 
        time=c( 1, 1, 2, 3, 4, 4, 5, 5, 8, 8, 8, 8, 11, 11, 12, 12, 15, 17, 22, 23 ), 
        age=c( 61, 65, 59, 52, 56, 67, 63, 58, 56, 58, 52, 49, 50, 55, 49, 62, 51, 49, 57, 52 ) 
    )
    DF$deceased = 1
    DF$deceased[10] = 0 # patient is still alive
    DF
}