pred_int: Function to calculate prediction interval

View source: R/pred_int.R

pred_intR Documentation

Function to calculate prediction interval

Description

Function to calculate prediction interval

Usage

pred_int(
  df,
  analysis_result = "analysis_result",
  distribution = "distribution",
  method = "Bonferroni",
  n_mean = 1,
  k = 1,
  pi_type = "upper",
  conf_level = 0.95
)

Arguments

df

df data frame of groundwater data in tidy format

analysis_result

the analysis result column

distribution

the distribution column

method

default is "Bonferroni"

n_mean

n.mean positive integer specifying the sample size associated with the future averages. The default value is n.mean=1 (i.e., individual observations). Note that all future averages must be based on the same sample size.

k

k positive integer specifying the number of future observations or averages the prediction interval should contain with confidence level conf.level. The default value is k=1.

pi_type

character string indicating what kind of prediction interval to compute. The possible values are pi_type="two-sided" (the default), pi_type="upper", and pi_type="lower".

conf_level

a scalar between 0 and 1 indicating the confidence level of the prediction interval. The default value is conf.level=0.95

Examples

data("gw_data")

wells <- c("MW-1", "MW-2", "MW-3", "MW-4")

params <- c("Sulfate, total",
            "Arsenic, dissolved",
            "Boron, dissolved")

background <- lubridate::ymd(c("2007-12-20", "2012-01-01"), tz = "UTC")

# first group data by location, param, and background
# estimate percent less than and distribution
background_data <- gw_data %>%
 filter(location_id %in% wells, param_name %in% params,
         sample_date >= background[1] & sample_date <= background[2]) %>%
  group_by(location_id, param_name, default_unit) %>%
  percent_lt() %>%
  est_dist(., keep_data_object = TRUE) %>%
  arrange(location_id, param_name)

background_data %>%
pred_int(., pi_type = "upper", conf_level = 0.99)



jentjr/gwstats documentation built on Jan. 12, 2024, 9:40 p.m.