Skip to contents

Evaluate outliers in plot format

Source: R/outlier_check.R
outlier_plot.Rd

Visualize spread of data and measures to identify outliers.

Usage

outlier_plot(
  dat,
  project,
  x,
  dat.remove = "none",
  sd_val = NULL,
  x.dist = "normal",
  date = NULL,
  group = NULL,
  pages = "single",
  output.screen = FALSE,
  log_fun = TRUE
)

Arguments

dat

Primary data containing information on hauls or trips. Table in the FishSET database contains the string 'MainDataTable'.

project

String, name of project.

x

Variable in dat to check for outliers.

dat.remove

Outlier measure. Values outside the measure are removed. Users can use the predefined values (see below) or user-defined distance from the mean. For user-defined values, dat.remove should be a numeric value. For example, dat.remove = 6 would would result in value outside 6SD from the mean being class as outliers. User-defined standard deviations from the mean can also be applied using sd_val. Pre-defined choices: "none", "5_95_quant", "25_75_quant", "mean_2SD", "median_2SD", "mean_3SD", "median_3SD". See the Details section for more information.

sd_val

Optional. Number of standard deviations from mean defining outliers. Example, sd_val = 6 would mean values outside +/- 6 SD from the mean would be outliers.

x.dist

Distribution of the data. Choices include: "normal", "lognormal", "exponential", "Weibull", "Poisson", "negative binomial".

date

(Optional) date variable to group the histogram by year.

group

(Optional) additional variable to group the histogram by.

pages

Whether to output plots on a single page ("single", the default) or multiple pages ("multi").

output.screen

Logical, if true, return plots to the screen. If FALSE, returns plot to the 'output' folder as a png file.

log_fun

Logical, whether to log function call (for internal use).

Value

Plot of the data

Details

The function returns three plots: the data, a probability plot, and a Q-Q plot. The data plot returns x against row number. Red points are data points that would be removed based on dat.remove. Blue points are data points within the bounds of dat.remove. If dat.remove is "none", then only blue points will be shown. The probability plot is a histogram of the data, after applying dat.remove, with the fitted probability distribution based on x.dist. group groups the histogram by a variable from dat, date groups the histogram by year. The Q-Q plot plots are sampled quantiles against theoretical quantiles, after applying dat.remove.

The dat.remove choices are:

  • numeric value: Remove data points outside +/- `x`SD of the mean

  • none: No data points are removed

  • 5_95_quant: Removes data points outside the 5th and 95th quantiles

  • 25_75_quant: Removes data points outside the 25th and 75th quantiles

  • mean_2SD: Removes data points outside +/- 2SD of the mean

  • median_2SD: Removes data points outside +/- 2SD of the median

  • mean_3SD: Removes data points outside +/- 3SD of the mean

  • median_3SD: Removes data points outside +/- 3SD of the median

The distribution choices are:

  • normal

  • lognormal

  • exponential

  • Weibull

  • Poisson

  • negative binomial

Examples

if (FALSE) {

outlier_plot(pollockMainDataTable, 'pollock', x = 'Haul', dat.remove = 'mean_2SD', 
             x.dist = 'normal', output.screen = TRUE)
# user-defined outlier        
outlier_plot(pollockMainDataTable, 'pollock', x = 'Haul', dat.remove = 6, 
             x.dist = 'lognormal', output.screen = TRUE)
}