Recreation of the animation of kernel estimation from "Visualizing How a Kernel Draws a Smooth Line" - https://www.walker-harrison.com/posts/2021-02-13-visualizing-how-a-kernel-draws-a-smooth-line/

Kernel estimation animation

## Code by Walker Harrison @walkwearscrocs
## From https://www.walker-harrison.com/posts/2021-02-13-visualizing-how-a-kernel-draws-a-smooth-line/

library(tidyverse)      

theme_set(theme_bw())
set.seed(0)

n <- 100
x <- runif(n, 0, 4*pi)
y <- sin(x)*x^2*exp(-x/2) + rnorm(n, sd = 0.3)
df <- data.frame(x, y)
x_new <- seq(min(x), max(x), length.out = 100)
y_smoothed <- map_dbl(x_new, ~sum(dnorm(x, mean = .x, sd = 0.5) * y)/
                        sum(dnorm(x, mean = .x, sd = 0.5)))
df_smoothed <- data.frame(x_new, y_smoothed)


##
## Recreation of the animation by June Choe @yjunechoe
##

library(gganimate)        # CRAN v1.0.7
library(ragg)             # CRAN v1.1.0

kernel_anim <- df_smoothed %>% 
  # Rename the `x_new` column to avoid ambiguity with the global variable `x_new`
  # because `transition_reveal()` evaluates the `along` argument in global env first
  rename(
    reveal_along = x_new,
    y = y_smoothed
  ) %>% 
  ggplot(aes(x = reveal_along, y = y)) +
  
  # Static global points
  geom_point(
    aes(x = x),
    data = df
  ) +
  
  # Dynamic local points
  geom_point(
    aes(x = x, size = size, col = col, group = group),
    data = map_dfr(df_smoothed$x_new, ~ {
      df %>% 
        mutate(
          reveal_along = .x,
          size = dnorm(x, mean = .x, sd = 0.5),
          col = dnorm(x, mean = .x, sd = 0.5),
          group = row_number()
        )
    }),
    show.legend = FALSE
  ) +
  
  # Moving line and point
  geom_line() +
  geom_point(size = 3, fill = "white", color = "black", shape = 21) +
  
  # Engine
  transition_reveal(reveal_along)


animate(
  kernel_anim, nframes = length(x_new), fps = 10,
  device = agg_png(), width = 6, height = 4, units = "in" , res = 300
)