create_statistics: Calculate statistics for the given dataset

In inbo/forrescalc: Calculation of Aggregated Values on Dendrometry, Regeneration and Vegetation of Forests, Starting from Individual Tree Measures from Fieldmap

Calculate statistics for the given dataset

Description

This function calculates statistics for the given data (e.g. from the git-repository forresdat) on the specified level (e.g. forest_reserve, period and species) and for the specified variables (e.g. basal_area and volume). Calculated statistics include number of observations, mean, variance and confidence interval with lower and upper limit (lci and uci).

These summary statistics are calculated on the given data, not taking into account absence of observations unless explicitly added as a record with value zero. E.g. if a certain species only occurs in 3 plots out of 10 and no records are added for the 7 remaining plots, the summary statistics (e.g. mean coverage) are calculated on 3 plots. Records with value zero for certain variables (e.g. coverage of a certain species or number of trees for a certain diameter class) can automatically be added using the function add_zeros().

In case of intervals, the variance and confidence interval are calculated based on the minimum and maximum values of the intervals of the individual records (which is considered a CI, so lci and uci can serve as min and max). For this, dataset must contain columns with minimum and maximum values, variables must contain a name for the output of this variable, and interval_information must contain the variable names for minimum, maximum and output that should be used. In interval_information it can be specified if a logarithmic transformation is needed to compensate of unequal interval widths. In this case, mean and the confidence interval are transformed back, but variance is not, as this result would be confusing rather than useful. For typical forresdat variables, the default value of interval_information can be used and in this case, the variable mentioned in variables should be named after the values in forresdat, omitting min_, ⁠_min⁠, max_ or ⁠_max⁠ (see example on interval data).

Usage

create_statistics(
  dataset,
  level = c("period", "forest_reserve"),
  variables,
  include_year_range = FALSE,
  na_rm = FALSE,
  interval_information = suppressMessages(read_csv2(system.file("extdata/class_data.csv",
    package = "forrescalc")))
)

Arguments

dataset	dataset with data to be summarised with at least columns year and period, e.g. table from git repository forresdat
level	grouping variables that determine on which level the values should be calculated (e.g. forest_reserve, year and species), given as a string or a vector of strings. Defaults to forest_reserve & period.
variables	variable(s) of which summary statistics should be calculated (given as a string or a vector of strings)
include_year_range	Should min_year and max_year be calculated based on a given column year in dataset? Defaults to FALSE.
na_rm	Should NA values in the dataset be ignored? Defaults to FALSE. If TRUE, levels without any non NA data are kept (resulting in NA values).
interval_information	overview of names for interval data, including columns var_name (= name for output), var_min and var_max (= names for minimum and maximum value in input dataset), and preferred_transformation (= "log" if log-transformation is desired). Defaults to a table containing all interval variables in forresdat, where log transformation is applied in variables where class widths differ. (In cover data in the Longo scale, log transformation is only applied in variables where most observations have a low coverage, e.g. moss cover, in congruence with the fact that class widths only differ in the lower part of the Longo scale.)

Value

dataframe with the columns chosen for level, a column variable with the chosen variables, and the columns n_obs, mean, variance, lci (lower limit of confidence interval) and uci (upper limit of confidence interval)

Examples

library(forrescalc)
dendro_by_plot <- read_forresdat_table(tablename = "dendro_by_plot")
create_statistics(
  dataset = dendro_by_plot,
  level = c("forest_reserve", "period"),
  variables = "vol_alive_m3_ha"
)
dendro_by_diam_plot_species <-
  read_forresdat_table(tablename = "dendro_by_diam_plot_species")
create_statistics(
  dataset = dendro_by_diam_plot_species,
  level = c("forest_reserve", "year", "species", "dbh_class_5cm"),
  variables = c("basal_area_alive_m2_ha", "basal_area_dead_m2_ha")
)
#example on interval data (shrub_cover and tree_cover)
veg_by_plot <- read_forresdat_table(tablename = "veg_by_plot")
create_statistics(dataset = veg_by_plot,
  level = c("forest_reserve", "period", "plottype"),
  variables = c("number_of_species", "shrub_cover", "tree_cover")
)
# example on data with confidence interval (number_established_ha and
# number_seedlings_ha)
reg_by_plot <-
  read_forresdat_table(tablename = "reg_by_plot")
create_statistics(dataset = reg_by_plot,
  level = c("forest_reserve", "period", "plot_id"),
  variables = c("number_established_ha", "number_seedlings_ha")
)

inbo/forrescalc documentation built on Sept. 28, 2024, 11:45 a.m.