Filter Histograms:

Small
Low fidelity
Cue to the user to lead them in narrowing a result set to find specific details

Bin size:

Narrow enough to reveal interesting features about the distribution
Wide enough to reduce noise
AREA matters, height is only 1 of the two relevant dimensions (bin widths can vary)
## Warning: `data_frame()` is deprecated, use `tibble()`.
## This warning is displayed once per session.

Naturalize Histogram by putting all outliers in one bin

This is raw data, with no manipulation.

Change Bin Width

This represents unmanipulated data where we have changed the binwidth

This is the same data, only we have bucketed anything over a specific amount (outliers)

Change Bin width and group outliers

This is an illustration where we have manipulated for outliers. We have also modified binwidth and it is in an overly wide state. Below, you will find a more readable sample where the binwidth is calculated based on the interquartile range and population size.

Calculation of Bin Width

Freedman-Diaconis

hist(hist_data$spanCount, breaks = "FD")

hist_data <-
  mutate(hist_data, spanCountNew = ifelse(hist_data$spanCount > 10, 10, hist_data$spanCount))
bw <- 2 * IQR(hist_data$spanCount) / length(hist_data$spanCount)^(1/3)
hist_data %>%
  ggplot(aes(spanCountNew)) +
  geom_histogram(binwidth = bw, col = "black", fill="blue")

hist(hist_data$spanCount)