gini-plot

Aufgabe

Visualisieren Sie die Gini-Funktion!

Lösung

library(tidyverse)

granularity <- .1
x1 = seq(from = 0, to = 1, by = granularity)
x2 = seq(from = 1, to = 0, by = -granularity)
#x2 <- 1 - x1
  
d <- expand_grid(x1, x2)

Gini-Loss:

gini_loss <- function(x1, x2) {1 - (x1^2 + x2^2)}

Funktion berechnen:

d2 <-
  d %>% 
  rowwise() %>% 
  mutate(y = gini_loss(x1, x2))

# d2 <-
#   outer(x1, x3, FUN = gini_loss) %>% 
#   as_tibble() %>% 
#   pivot_longer(cols = everything())

# d <-
#   d %>% 
#   mutate(
#     x3 = 1 - x1,
#     y = 1 - (x1^2 + x3^2))

d2 %>% 
  ggplot(aes(x1, x2, fill = y)) +
  geom_tile() +
  scale_x_continuous(limits = c(-2, 2)) +
  scale_y_continuous(limits = c(-2, 2))

So sieht der Funktionsgraph in Geogebra aus.

---

Categories:

• 2023
• vis
• statlearning
• tree
• string