ggplot2 dream functions
Document functions I have no real intention of implementing but would love to see in ggplot2 or some extension.
2D-density for sfc POINT objects. Output can be one of point density or contoured lines.
# Data has units of [m] (meter)
data %>%
ggplot() +
geom_sf_density(aes(fill = stat(level)),
grid.cell = c(1e3, 1e3),
geom = "polygon",
contour = FALSE) +
scale_fill_viridis()Bin counts for sfc POINT objects.
Need to read more into variogram modelling before a current and idealized implementation can be fully realized.
library(ggplot2)
library(dplyr)
library(gstat)
# Current implementation ----------------------------------------------
# WIP
# Idealized implementation --------------------------------------------
# WIP
df %>%
ggplot() +
geom_sf_krige(aes(formula = x ~ y))Inspired by D3's curve API. Allows user to specify interpolation method (spline, step, step-before, step-after, etc) and closure (do start and end points connect).
library(ggplot2)
library(dplyr)
# Current implementation ----------------------------------------------
# WIP
# Idealized implementation --------------------------------------------
# WIP
df %>%
ggplot(aes(x, y)) +
geom_line_shape(method = "step", args = list(when = "after"))
# -- OR --
df %>%
ggplot(aes(x, y)) +
geom_line_shape(stat = "step_after")
# default stat is "step_middle"
df %>%
ggplot(aes(x, y)) +
geom_line_shape()
# connect controls whether to connect start and end point
df %>%
ggplot(aes(x, y)) +
geom_line_shape(stat = "xspline", connect = TRUE)Breaks a time series chart into multiple facets (one column, many rows).
library(ggplot2)
library(dplyr)
# Current implementation ----------------------------------------------
txhousing %>%
filter(city == "Abilene") %>%
mutate(date = paste(year, month, "1", sep = "-"),
date = as.Date(date),
rbreak = ceiling(seq(1, 187) / 34)) %>%
ggplot(aes(date, sales)) +
geom_line() +
facet_wrap(vars(rbreak), scales = "free_x", ncol = 1)
# Idealized implementation --------------------------------------------
txhousing %>%
ggplot(aes(date, sales)) +
geom_line() +
facet_break(vars(date), breaks = 6)
Performs kmeans clustering on two variables; geom_point() compliments this facet.
# Current Implementation ----------------------------------------------
# Number of clusters
n <- 2:4
kdata <- iris[, c("Sepal.Length", "Petal.Length")]
# Run kmeans
klusters <- lapply(n, kmeans, x = kdata)
# Extract clusters
clusters <- Reduce(cbind, lapply(klusters, `[[`, "cluster"))
colnames(clusters) <- paste0("k", n)
kdata_long <- cbind(kdata, clusters) %>%
tidyr::gather(... = k2:k4)
kdata_long %>%
ggplot(aes(x = Sepal.Length,
y = Petal.Length,
color = factor(value))) +
geom_point() +
facet_wrap(vars(key))
# Idealized Implementation --------------------------------------------
iris %>%
ggplot(aes(x = Sepal.Length, y = Petal.Length)) +
geom_point() +
facet_kmeans(
k = 2:4, # passed on to kmeans()
aesthetics = c("color") # or "fill" if pch = 21
)
# Also facet_sf_kmeans()...wip
# Can also extend to dbscan::dbscan, cluster::pam, etc.