[关闭]
@fanxy 2020-03-15T08:52:43.000000Z 字数 8850 阅读 5859

第三讲 数据可视化与编程基础

樊潇彦 复旦大学经济学院 金融数据

0. 准备工作

下载 Ch03.rar，解压缩后存于工作目录下。 

  1. setwd("D:\\...\\Ch03")          # 设定工作目录，注意为/或\\
  2. rm(list=ls())                   # 清内存
  4. ## 调用之前已安装且当前要用的包
  5. library(tidyverse)
  6. library(readstata13)
  7. library(haven) 
  8. library(readxl)
  10. ## 安装和调用本节要用的包
  11. install.packages(c("ggplot2","ggvis","shiny","dygraphs"))
  12. library(ggplot2)
  13. library(ggvis)
  14. library(shiny)
  15. library(dygraphs)

1. 数据可视化

1.1 基础作图

  1. # 基础绘图 plot，包括 points, lines, scatter, histogram等
  2. data(iris)
  3. plot(iris$Sepal.Length)                                  # 单变量点图
  4. plot(iris$Sepal.Length,type="l")                         # 单变量线图
  5. plot(x=iris$Sepal.Length, y=iris$Petal.Length)           # 双变量散点图
  6. plot(x=iris$Sepal.Length, y=iris$Petal.Length, type="h") # 双变量柱状图
  7. plot(x=iris$Species, y=iris$Petal.Length, type="h")      # 连续变量按离散变量分组的箱式图
  9. # 条形图 barplot
  10. msft <- c(26.85,27.41,28.21,32.64,34.66,34.30,31.62,33.40)
  11. msft.returns <- msft[-1] / msft[-length(msft)] - 1
  12. names(msft.returns) <- month.abb[1:length(msft.returns)]
  13. barplot(msft.returns, col="blue")
  14. barplot(msft.returns, names.arg=month.name[1:length(msft.returns)],
  15.   col="blue",las=2)                   
  17. # 饼图 pie
  18. x <-1:5;pie(x,col=rainbow(5))
  19. box()
  21. # 向日葵图 sunflowerplot
  22. sunflowerplot(iris[,3:4])
  24. # 绘制矩阵或数据框的二元图 pairs
  25. data(iris)
  26. pairs(iris[1:4], main = "Anderson's Iris Data -- 3 species",pch = 21, 
  27.       bg = c("red", "green3", "blue")[unclass(iris$Species)])
  29. # 多个数据作图 matplot
  30. set.seed(1)
  31. x <- cumsum(rnorm(50))
  32. y <- cumsum(rnorm(50))
  33. z <- cumsum(rnorm(50))
  34. matplot(cbind(x,y,z),col=2:4,type="l",lty=1, xlab="", ylab="")
  35. legend("bottom",legend=c("x","y","z"),
  36.        lty=1,col=2:4,bty="n")
  38. # 根据指定函数绘制指定范围的曲线图
  39. curve(sin, -2*pi, 2*pi, xname = "t")

1.2 修饰和保存

  1. # 以参数`par`和标签`legend`为例：
  2. intc <- c(20.42,20.48,21.43,23.50,24.04,24.00,23.11,21.98)
  3. intc.returns <- intc[-1] / intc[-length(intc)] - 1
  4. barplot(rbind(msft.returns,intc.returns),beside=T,col=c(2,4))
  5. legend(x="topleft",legend=c("MSFT","INTC"),pch=15,col=c(2,4),bty="n")
  7. # 生成一个绘图窗口在其中绘制图形后用savePlot()函数保存
  8. windows()
  9. plot(1:10)
  10. rect(1, 5, 3, 7, col="blue")
  11. savePlot("test01", type="jpg",device=dev.cur(),restoreConsole=TRUE)
  13. # 直接在jpeg设备上绘制图形，完成后使用dev.off()关闭设备，存盘退出
  14. jpeg(file="myplot.jpeg")
  15. plot(1:10)
  16. rect(1, 5, 3, 7, col="blue")
  17. dev.off()

1.3 ggplot2

Basic Components of a ggplot2 Plot

ggplot_command.jpg-141.5kB

  1. library(ggplot2)
  2. data(diamonds)    # 调用钻石数据 
  3. set.seed(42)      # 设随机数，抽1000个样本
  4. small <- diamonds[sample(nrow(diamonds), 1000), ] 
  5. head(small)
  7. # 生成空白图，查看ggplot对象要素
  8. p = ggplot()
  9. class(p)
  10. names(p)
  12. # 数据和映射
  13. # 以克拉(carat)、切工(cut)、透明度(clarity)等因素对钻石价格(price)的影响为例：
  14. ggplot(data = small, 
  15.        mapping = aes(x = carat, y = price,  # 纵横轴
  16.                           color = cut, shape = clarity)
  17.        ) +
  18.   geom_point()  
  20. # 几何与统计
  21. ggplot(small) +
  22.   geom_density(aes(x=price, colour=cut))
  24. # 分面与标签
  25. ggplot(small, aes(x=carat, y=price, color=color))+      # 标题、纵横轴标签
  26.   geom_point() +
  27.   facet_wrap(~cut) +                                    # 一页多图
  28.   labs(title ="Diamonds", x = "Carat", y = "Price") +
  29.   theme_bw()

1.4 ggvis

1.主要命令

  1. library(ggvis)
  2. library(dplyr)
  3. data(mtcars)
  5. # 散点和拟合线：points, smooths
  6. mtcars %>%
  7.   ggvis(~wt, ~mpg) %>%
  8.   layer_points(fill= ~factor(cyl))%>%   # 除填充(fill)外，还可设边界(stroke)、大小(size)、形状(shape)和透明度(opacity)
  9.   layer_lines(stroke:= "gray") %>%     
  10.   layer_smooths()
  12. # 线和条：lines, bars
  13. data(pressure)
  14. pressure %>% 
  15.   ggvis(~temperature, ~pressure) %>%
  16.   layer_lines(stroke := "red") %>%
  17.   layer_bars(width=20, fill:=NA)

2.统计作图

  1. # Boxplots
  2. mtcars %>% 
  3.   ggvis(~factor(cyl), ~mpg) %>% 
  4.   layer_boxplots()
  6. # Histograms
  7. cocaine %>% 
  8.   ggvis(~potency) %>%
  9.   layer_histograms(width = 10, center = 0, fill := "pink") %>%
  10.   add_axis("x", title = "potency") %>%
  11.   add_axis("y", title = "histograms")
  13. # Densities
  14. cocaine %>% 
  15.   ggvis(~potency) %>%
  16.   layer_densities(fill := NA, stroke := "red") %>%
  17.   add_axis("x", title = "potency") %>%
  18.   add_axis("y", title = "densities")

3.回归预测

  1. mtcars %>% 
  2.   ggvis(~wt, ~mpg) %>%
  3.   layer_points() %>%
  4.   layer_smooths() %>%
  5.   layer_model_predictions(stroke := "red", model = "lm", se = TRUE)
  7. mtcars %>% 
  8.   ggvis(~wt, ~mpg, fill = ~factor(cyl)) %>% 
  9.   layer_points() %>% 
  10.   group_by(cyl) %>%                           # 分组回归
  11.   layer_model_predictions(model = "lm")

4.交互式作图

  1. # 1) 连续调整
  2. mtcars %>%
  3.   ggvis(~wt, ~mpg) %>%
  4.   layer_points(size := input_slider(10, 100, value = 30))%>%  # 调节点大小
  5.   layer_smooths(span = input_slider(0.5, 1, value = 1))       # 调节拟合的窗宽
  7. # 2) 文本框
  8. mtcars %>% 
  9.   ggvis(x = ~wt) %>% 
  10.   layer_densities(
  11.     adjust = input_slider(.1, 2, value = 1, step = .1, label = "Bandwidth adjustment"),
  12.     kernel = input_select(c("Gaussian" = "gaussian", 
  13.                              "Epanechnikov" = "epanechnikov",
  14.                              "Rectangular" = "rectangular", 
  15.                              "Triangular" = "triangular", 
  16.                              "Biweight" = "biweight", 
  17.                              "Cosine" = "cosine", 
  18.                              "Optcosine" = "optcosine"), 
  19.                            label = "Kernel")) 
  20. # 3) 动态图
  21. dat <- data.frame(time = 1:10, value = runif(10))
  22. ddat <- reactive({invalidateLater(2000, NULL)
  23.   dat$time <<- c(dat$time[-1], dat$time[length(dat$time)] + 1)
  24.   dat$value <<- c(dat$value[-1], runif(1))
  25.   dat  })
  26. ddat %>% 
  27.   ggvis(x = ~time, y = ~value, key := ~time) %>%
  28.   layer_points() %>%
  29.   layer_paths()

1.5 dygraphs

  1. library(dygraphs) 
  2. # dyAnnotation：标注
  3. dygraph(presidents, main = "Presidential Approval") %>%
  4.   dyAxis("y", valueRange = c(0, 100)) %>%
  5.   dyAnnotation("1950-7-1", text = "A", tooltip = "Korea") %>%
  6.   dyAnnotation("1965-1-1", text = "B", tooltip = "Vietnam")
  8. # dyAxis：坐标轴
  9. dygraph(nhtemp, main = "New Haven Temperatures") %>%
  10.   dyAxis("y", label = "Temp (F)", valueRange = c(40, 60)) %>%
  11.   dyOptions(axisLineWidth = 1.5, fillGraph = TRUE, drawGrid = FALSE)
  13. # dyEvent：事件
  14. dygraph(presidents, main = "Presidential Approval") %>%
  15.   dyAxis("y", valueRange = c(0, 100)) %>%
  16.   dyEvent("1950-6-30", "Korea", labelLoc = "bottom") %>%
  17.   dyEvent("1965-2-09", "Vietnam", labelLoc = "bottom")
  19. # dyHighlight：提亮
  20. lungDeaths <- cbind(ldeaths, mdeaths, fdeaths)
  21. dygraph(lungDeaths, main = "Deaths from Lung Disease (UK)") %>%
  22.   dyHighlight(highlightCircleSize = 5,
  23.               highlightSeriesBackgroundAlpha = 0.2,
  24.               hideOnMouseOut = FALSE)
  26. # dyLegend：标签
  27. dygraph(nhtemp, main = "New Haven Temperatures") %>%
  28.   dySeries("V1", label = "Temperature (F)") %>%
  29.   dyLegend(show = "always", hideOnMouseOut = FALSE)
  31. # dyLimit：极值
  32. dygraph(presidents, main = "Presidential Approval") %>%
  33.   dyAxis("y", valueRange = c(0, 100)) %>%
  34.   dyLimit(max(presidents, na.rm = TRUE), "Max",
  35.           strokePattern = "solid", color = "blue")
  37. # dyOptions：选项
  38. dygraph(lungDeaths) %>% dyRangeSelector()
  39. dygraph(lungDeaths) %>%
  40.   dySeries("mdeaths", label = "Male") %>%
  41.   dySeries("fdeaths", label = "Female") %>%
  42.   dyOptions(stackedGraph = TRUE) %>%
  43.   dyRangeSelector(height = 20)
  45. hw <- HoltWinters(ldeaths)
  46. predicted <- predict(hw, n.ahead = 72, prediction.interval = TRUE)
  47. dygraph(predicted, main = "Predicted Lung Deaths (UK)") %>%
  48.   dyAxis("x", drawGrid = FALSE) %>%
  49.   dySeries(c("lwr", "fit", "upr"), label = "Deaths") %>%
  50.   dyOptions(colors = RColorBrewer::brewer.pal(3, "Set1"))
  52. # dyRangeSelector：时间区
  53. dygraph(nhtemp, main = "New Haven Temperatures") %>%
  54.   dyRangeSelector()
  55. dygraph(nhtemp, main = "New Haven Temperatures") %>%
  56.   dyRangeSelector(dateWindow = c("1920-01-01", "1960-01-01"))
  57. dygraph(nhtemp, main = "New Haven Temperatures") %>%
  58.   dyRangeSelector(height = 20, strokeColor = "")
  60. # dyRoller：滚动平滑
  61. # Y values are averaged over the specified number of time scale units.
  62. dygraph(discoveries, main = "Important Discoveries") %>%
  63.   dyRoller(rollPeriod = 5)
  65. # dyShading：阴影区
  66. dygraph(nhtemp, main = "New Haven Temperatures") %>%
  67.   dyShading(from = "1920-1-1", to = "1930-1-1") %>%
  68.   dyShading(from = "1940-1-1", to = "1950-1-1")
  69. dygraph(nhtemp, main = "New Haven Temperatures") %>%
  70.   dyShading(from = "48", to = "52", axis = "y") %>%
  71.   dyShading(from = "50", to = "50.1", axis = "y", color = "black")

2. 编程基础

介绍变量赋值、分支结构、循环结构、函数使用、获取帮助等知识。

2.1 通过赋值生成一个新变量

  1. x  <- 1.5
  2. cat("x = ",x,"\n",sep="")    # 屏幕显示，也可用于测试程序
  3. y1 <- c(1.5,2.3,8.6,7.4,9.2)
  4. y2 <- c("MSFT","GOOG","AAPL")
  5. y3 <- c(T,F,T,T,F,F)
  6. 3.1415926 -> z;              # 数据在左，变量名在右赋值，但比较少用
  7. assign("t",1.414)            # assign()函数给变量赋值

2.2 分支结构：if, if-else

  1. # if
  2. a <- 1
  3. if(a==1) print("a==1")
  4. a <- 2
  5. if(a > 1)  print("a > 1")   else   print("a <= 1")
  6. a <- 3
  8. if( a == 1){ 
  9.   a          # 不会显示 a 的值
  10.   print("I am a boy!")
  11.   }else{     # 如果有多行命令，需要用{}引起来，else必须紧跟在}后面
  12.   print(a)   # 会显示 a 的值
  13.   print("I am a girl!")  
  14.   }

2.3 多重分支结构：if, ifelse, switch

  1. # 1) if - else if
  2. a <- 4
  3. if( a == 1)
  4. {
  5.    print("a == 1")
  6. }else if( a == 2)   # 同样每个else必须和前面的}紧紧粘在一起
  7. {
  8.    print("a == 2")
  9. }else
  10. {
  11.   print("Not 1 & 2")
  12. }
  14. # 2) ifelse()计算第一个逻辑表达式得到结果如果为T则返回第二个参数；否则返回第三个参数
  15. a <- 2
  16. ifelse(a > 1,3.1416,1.414) 
  18. # 3) switch语句的多重分支结构
  19. switch(a,
  20.        print("选项1"),
  21.        print("选项2"),
  22.        print("选项3"),
  23.        print("选项4"),
  24.        print("选项5")
  25.        )

2.4 循环结构: for, while, repeat

  1. # 1) for
  2. iTotal <-  0
  3. for(i in 1:100)                # 用关键词in枚举向量中的每一整数
  4. {
  5.    iTotal <- iTotal + i
  6. } 
  7. cat("1-100的累加和为：",iTotal,"\n",sep="")
  9. szSymbols <- c("MSFT","GOOG","AAPL","INTL","ORCL","SYMC")
  10. for(SymbolName in szSymbols)           # 字符串也可以枚举
  11. {
  12.    cat(SymbolName,"\n",sep="")
  13. }
  15. # 2) while
  16. i <- 1
  17. iTotal <- 0
  18. while(i <= 100)
  19. {
  20.    iTotal <- iTotal + i
  21.    i <- i + 1
  22. }
  23. cat("1-100的累加和为：",iTotal,"\n",sep="")  # 屏幕显示结果
  25. # 3) repeat
  26. i <- 1
  27. iTotal <- 0
  28. repeat                                       # 无条件循环，必须在程序内部设法退出
  29. {
  30.   iTotal <- iTotal + i
  31.   i <- i + 1
  33.   if(i <= 100) next else break               # 注意：next,break的用法
  34. }
  35. cat("1-100的累加和为：",iTotal,"\n",sep="")

2.5 自定义函数 function

  1. # 对于小函数，可写好后直接调用。如计算矩阵的幂：
  2. mat_power = function(A, n){
  3.   Apower=A
  4.   for (i in 2:n) Apower= Apower %*% A
  5.   return(Apower)
  6. }
  8. A = matrix(c(1:4),2)
  9. mat_power(A, 3)
  10. A %*% A %*% A
  12. # 对于较大的函数，要另存为.r 文件，再调用。
  13. rm("mat_power")
  14. source("myfun.r")        # 调用自编程序
  15. mat_power(A, 3)

2.6 获取帮助信息

  1. ?print                # 在RStudio右侧打开相关帮助界面
  2. example(print)        # 命令示例
  3. ?quantmod             # 打开扩展包整体帮助信息
  4. apropos("print*")     # 在搜索路径下查找满足正则表达式的所有函数信息
  5. demo(graphics)
  6. # 如果对包或命令的具体名称不清楚，可以从 google 或 http://rseek.org/ 上查找。
  1. R.I. Kabacoff著：《R语言实战（第2版）》，王小宁、刘撷芯、黄俊文译，人民邮电出版社，2016
  2. Rstudio: ggplot2 Cheat Sheet
  3. Roger D. Peng：Exploratory Data Analysis, Lecture Notes
  4. The DataCamp Team: Questions All R Users Have About Plots
添加新批注
在作者公开此批注前，只有你和作者可见。
回复批注