Diagnostic Distribution Plots

Diagnostic distribution plots: poissonness, binomialness and negative binomialness plots.

distplot(x, type = c("poisson", "binomial", "nbinomial"),
  size = NULL, lambda = NULL, legend = TRUE, xlim = NULL, ylim = NULL,
  conf_int = TRUE, conf_level = 0.95, main = NULL,
  xlab = "Number of occurrences", ylab = "Distribution metameter",
  gp = gpar(cex = 0.8), lwd=2, gp_conf_int = gpar(lty = 2),
  name = "distplot", newpage = TRUE,
  pop =TRUE, return_grob = FALSE, ...)

Arguments

x: either a vector of counts, a 1-way table of frequencies of counts or a data frame or matrix with frequencies in the first column and the corresponding counts in the second column.
type: a character string indicating the distribution.
size: the size argument for the binomial and negative binomial distribution. If set to NULL and type is "binomial", then size is taken to be the maximum count. If set to NULL and type is "nbinomial", then size is estimated from the data.
lambda: parameter of the poisson distribution. If type is "poisson" and lambda is specified a leveled poissonness plot is produced.
legend: logical. Should a legend be plotted?
xlim: limits for the x axis.
ylim: limits for the y axis.
conf_int: logical. Should confidence intervals be plotted?
conf_level: confidence level for confidence intervals.
main: a title for the plot.
xlab: a label for the x axis.
ylab: a label for the y axis.
gp: a "gpar" object controlling the grid graphical parameters of the points.
gp_conf_int: a "gpar" object controlling the grid graphical parameters of the confidence intervals.
lwd: line width for the fitted line
name: name of the plotting viewport.
newpage: logical. Should grid.newpage be called before plotting?
pop: logical. Should the viewport created be popped?
return_grob: logical. Should a snapshot of the display be returned as a grid grob?
...: further arguments passed to grid.points.

Details

distplot plots the number of occurrences (counts) against the distribution metameter of the specified distribution. If the distribution fits the data, the plot should show a straight line. See Friendly (2000) for details.

In these plots, the open points show the observed count metameters; the filled points show the confidence interval centers, and the dashed lines show the conf_level confidence intervals for each point.

Value

Returns invisibly a data frame containing the counts (Counts), frequencies (Freq) and other details of the computations used to construct the plot.

Author

Achim Zeileis Achim.Zeileis@R-project.org

References

D. C. Hoaglin (1980), A poissonness plot, The American Statistican, 34, 146–149.

D. C. Hoaglin & J. W. Tukey (1985), Checking the shape of discrete distributions. In D. C. Hoaglin, F. Mosteller, J. W. Tukey (eds.), Exploring Data Tables, Trends and Shapes, chapter 9. John Wiley & Sons, New York.

M. Friendly (2000), Visualizing Categorical Data. SAS Institute, Cary, NC.

Examples

## Simulated data examples:
dummy <- rnbinom(1000, size = 1.5, prob = 0.8)
distplot(dummy, type = "nbinomial")


## Real data examples:
data("HorseKicks")
data("Federalist")
data("Saxony")
distplot(HorseKicks, type = "poisson")

distplot(HorseKicks, type = "poisson", lambda = 0.61)

distplot(Federalist, type = "poisson")

distplot(Federalist, type = "nbinomial", size = 1)

distplot(Federalist, type = "nbinomial")

distplot(Saxony, type = "binomial", size = 12)