ggfortify包学习

library(ggfortify)
library(tidyverse)

#PCA (主成分分析)
df <- iris[c(1, 2, 3, 4)]
prcomp(df) %>%
  #autoplot()
  #选择数据中的一列来给画出的点按类别自动分颜色
  #autoplot(data = iris, colour = 'Species')
  #给定label = TRUE 可以给每个点加上标识(以rownames为标准)，也可以调整标识的大小
  #autoplot( data = iris, colour = 'Species', label = TRUE, label.size = 3)
  #给定 shape = FALSE 可以让所有的点消失，只留下标识，这样可以让图更清晰，辨识度更大
  # autoplot(data = iris, colour = 'Species', shape = FALSE, label.size = 3)
  # 给定 loadings = TRUE 可以很快的画出特征向量。
  # autoplot(data = iris, colour = 'Species', loadings = TRUE)
  # 可以显示特征向量的标识以及调整他们的大小
  autoplot(data = iris, colour = 'Species',
           loadings = TRUE, loadings.colour = 'blue',
           loadings.label = TRUE, loadings.label.size = 3)

# 因素分析
factanal(state.x77, factors = 3, scores = 'regression') %>%
  #autoplot(data = state.x77, colour = 'Income')
  autoplot(label = TRUE, label.size = 3,
         loadings = TRUE, loadings.label = TRUE, loadings.label.size  = 3)

#K-均值
kmeans(USArrests, 3) %>%
  # autoplot(data = USArrests)
  autoplot(data = USArrests, label = TRUE, label.size = 3)

#cluster(集群)
library(cluster)
clara(iris[-5],3) %>%
  # autoplot()
  #给定 frame = TRUE，可以把stats::kmeans 和 cluster::* 的每个集群圈出来。
  # autoplot(frame=3)
  # 可以通过 frame.type 来选择圈的类型
  autoplot(frame = TRUE, frame.type = 'norm')

# lfda(Fisher局部判别分析)
library(lfda)
# Fisher局部判别分析 (LFDA)
lfda(iris[-5], iris[, 5], 4, metric="plain") %>%
  # autoplot(data = iris, frame = TRUE, frame.colour = 'Species')
  # 半监督Fisher局部判别分析 (SELF)
  autoplot(data = iris, frame = TRUE, frame.colour = 'Species')


#时间序列的可视化
AirPassengers %>%
  #autoplot()
# 使用 ts.colour 和 ts.linetype来改变线的颜色和形状
autoplot(ts.colour = 'red', ts.linetype = 'dashed')

#多变量时间序列
library(vars)
Canada %>%
  # autoplot()
  # facets = FALSE 可以把所有变量画在一条轴上
  autoplot(facets = FALSE)

autoplot(AirPassengers, ts.colour = ('dodgerblue3'))
# 通过 ts.geom 来改变几何形状，目前支持的有 line, bar and point
autoplot(AirPassengers, ts.geom = 'bar', fill = 'blue')
autoplot(AirPassengers, ts.geom = 'point', shape = 3)

# forecast包
library(forecast)
d.arima <- auto.arima(AirPassengers)
forecast(d.arima, level = c(95), h = 50) %>%
  autoplot()

# vars包
library(vars)
d.vselect <- VARselect(Canada, lag.max = 5, type = 'const')$selection[1]
d.var <- VAR(Canada, p = d.vselect, type = 'const')
autoplot(predict(d.var, n.ahead = 50), ts.colour = 'dodgerblue4',
         predict.colour = 'blue', predict.linetype = 'dashed')


# changepoint包
library(changepoint)
autoplot(cpt.meanvar(AirPassengers))

library(strucchange)
autoplot(breakpoints(Nile ~ 1))


# https://terrytangyuan.github.io/2015/11/24/ggfortify-intro/

# Draw subplots
res <- lm(Volume ~ Girth, data = trees) 
autoplot(res, ncol = 4)+
  theme_classic()
autoplot(res, which = 1:6, ncol = 2, label.size = 3)

m <- lm(Petal.Width ~ Petal.Length, data = iris)
# Change the color by groups (species)
autoplot(m, which = 1:6, label.size = 3, data = iris,
         colour = 'Species')

# Compute a generalized linear model
m <- glm(Murder ~ Assault + UrbanPop + Rape,
         family = gaussian, data = USArrests)
# Create the plot
# Change the theme and colour
autoplot(m, which = 1:6, ncol = 2, label.size = 3,
         colour = "steelblue") + theme_bw()

# Draw multiple instances
library(purrr)
res <- purrr::map(c(3, 4, 5), ~ kmeans(iris[-5], .))
autoplot(res, data = iris[-5], ncol = 3)

library(survival)
res <- list(a = survfit(Surv(time, status) ~ 1, data = lung),
            b = survfit(Surv(time, status) ~ sex, data = lung))
autoplot(res)

# Plotting matrix
df <- mtcars[, c("mpg", "disp", "hp", "drat", "wt")] %>%
  as.matrix()
# Heatmap
autoplot(scale(df))


df2 <- df[, c("wt", "mpg")]
colnames(df2) <- c("V1", "V2")
# Scatter plot
autoplot(df2, geom = 'point') +
  labs(x = "mpg", y = "wt")