bin: Bin

In vbonhomme/pataqu: Permutationnal Analysis of Terminus Ante and post QUem data

Bin

Description

Bin and summarise without fitting

Usage

bin(df, y = y, x = x_new, by = NULL, fun = stats::median, x_bin, k = k)

Arguments

df	tibble() typically returned by quake()
y, x	colnames to use. Default to y/x_new
by	colname for grouping structure (besides k). Default to NULL
fun	to use to summarise y
x_bin	breaks as numeric vector, passed to cut(). You can use one of the cutter to define them. See Examples
k	colname for iteration . Default to k.

Examples

# mean, per taxa, per iteration and custom x_bin range
animals_q %>% bin(y=value, by=taxa, fun=mean, x_bin=seq(-100, 100, 10))

# shall you need help for binning, you can use one of cutters
## 10 groups of equal range
cutter_interval(animals_q, n=10, x_new)
## 7 groups with ~same number of observations
cutter_number(animals_q, n=10, x_new)
## bin every 100 years, note that we use tpq and taq 'original' columns here
cutter_width(animals_q, width=100, tpq, taq)

# then wrap the result you like with cutter_to_seq
# to go back to numbers and pass the result to bin eg:
equal_n <- cutter_to_seq(cutter_number(animals_q, n=10, x_new))
animals_equal_n <- bin(animals_q, y=value, by=taxa,
                       fun=mean, x_bin=equal_n)
# let's check that groups have roughly the same size:
dplyr::count(animals_equal_n, x_bin) # sounds good

# you can use other functions eg median,
# and do not specify by so that all taxa are merged
# and a global median is returned
bin(animals_q, y=value, x_bin=equal_n, fun=median)

# if you want many functions at once, you do not need bin
# when you have dplyr. See ?summarise
animals_q %>%
  dplyr::mutate(x_bin=cut(x_new, breaks=equal_n)) %>%
  dplyr::group_by(k, taxa, x_bin) %>%
  dplyr::summarise(
    qs    = stats::quantile(value, c(0.1, 0.25, 0.75, 0.9)), prob = c(0.1, 0.25, 0.75, 0.9),
    var   = stats::var(value),
    range = max(value)-min(value), .groups = "drop") %>%
    # then a little pivot_wider to get all quantile columns
 tidyr::pivot_wider(values_from = qs,
                    names_from = prob,
                    names_prefix="q")

vbonhomme/pataqu documentation built on April 24, 2022, 10:03 p.m.